Skip to content

Common principles


The BIDS specification is written in American English.


The keywords "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].

Based on these keywords, we define three requirement levels for specifying data or metadata: REQUIRED, RECOMMENDED, and OPTIONAL. The guiding principles for when particular data is placed under a given requirement level can be loosely described as below:

  • REQUIRED: Data cannot be be interpreted without this information (or the ambiguity is unacceptably high)
  • RECOMMENDED: Interpretation/utility would be dramatically improved with this information
  • OPTIONAL: Users and/or tools might find it useful to have this information

Throughout this specification we use a list of terms and abbreviations. To avoid misunderstanding we clarify them here.

  1. Dataset - A set of neuroimaging and behavioral data acquired for a purpose of a particular study. A dataset consists of data acquired from one or more subjects, possibly from multiple sessions.

  2. Modality - The category of brain data recorded by a file. For MRI data, different pulse sequences are considered distinct modalities, such as T1w, bold or dwi. For passive recording techniques, such as EEG, MEG or iEEG, the technique is sufficiently uniform to define the modalities eeg, meg and ieeg. When applicable, the modality is indicated in the suffix. The modality may overlap with, but should not be confused with the data type.

  3. Data type - A functional group of different types of data. Data files are contained in a directory named for the data type. In raw datasets, the data type directory is nested inside subject and (optionally) session directories. BIDS defines the following data types:

    1. func (task based and resting state functional MRI)

    2. dwi (diffusion weighted imaging)

    3. fmap (field inhomogeneity mapping data such as field maps)

    4. anat (structural imaging such as T1, T2, PD, and so on)

    5. perf (perfusion)

    6. meg (magnetoencephalography)

    7. eeg (electroencephalography)

    8. ieeg (intracranial electroencephalography)

    9. beh (behavioral)

    10. pet (positron emission tomography)

    11. micr (microscopy)

    12. nirs (near infrared spectroscopy)

    13. motion (motion)

  4. Subject - A person or animal participating in the study. Used interchangeably with term Participant.

  5. Session - A logical grouping of neuroimaging and behavioral data consistent across subjects. Session can (but doesn't have to) be synonymous to a visit in a longitudinal study. In general, subjects will stay in the scanner during one session. However, for example, if a subject has to leave the scanner room and then be re-positioned on the scanner bed, the set of MRI acquisitions will still be considered as a session and match sessions acquired in other subjects. Similarly, in situations where different data types are obtained over several visits (for example fMRI on one day followed by DWI the day after) those can be grouped in one session. Defining multiple sessions is appropriate when several identical or similar data acquisitions are planned and performed on all -or most- subjects, often in the case of some intervention between sessions (for example, training). In the PET context, a session may also indicate a group of related scans, taken in one or more visits.

  6. Sample - A sample pertaining to a subject such as tissue, primary cell or cell-free sample. Sample labels MUST be unique within a subject and it is RECOMMENDED that they be unique throughout the dataset.

  7. Data acquisition - A continuous uninterrupted block of time during which a brain scanning instrument was acquiring data according to particular scanning sequence/protocol.

  8. Task - A set of structured activities performed by the participant. Tasks are usually accompanied by stimuli and responses, and can greatly vary in complexity. For the purpose of this specification we consider the so-called "resting state" a task. In the context of brain scanning, a task is always tied to one data acquisition. Therefore, even if during one acquisition the subject performed multiple conceptually different behaviors (with different sets of instructions) they will be considered one (combined) task.

  9. Event - Something that happens or may be perceived by a test subject as happening at a particular instant during the recording. Events are most commonly associated with on- or offset of stimulus presentations, or with the distinct marker of on- or offset of a subject's response or motor action. Other events may include unplanned incidents (for example, sudden onset of noise and vibrations due to construction work, laboratory device malfunction), changes in task instructions (for example, switching the response hand), or experiment control parameters (for example, changing the stimulus presentation rate over experimental blocks), and noted data feature occurrences (for example, a recording electrode producing noise). In BIDS, each event has an onset time and duration. Note that not all tasks will have recorded events (for example, "resting state").

  10. Run - An uninterrupted repetition of data acquisition that has the same acquisition parameters and task (however events can change from run to run due to different subject response or randomized nature of the stimuli). Run is a synonym of a data acquisition. Note that "uninterrupted" may look different by modality due to the nature of the recording. For example, in MRI or MEG, if a subject leaves the scanner, the acquisition must be restarted. For some types of PET acquisitions, a subject may leave and re-enter the scanner without interrupting the scan.

  11. index - A nonnegative integer, possibly prefixed with arbitrary number of 0s for consistent indentation, for example, it is 01 in run-01 following run-<index> specification.

  12. label - An alphanumeric value, possibly prefixed with arbitrary number of 0s for consistent indentation, for example, it is rest in task-rest following task-<label> specification. Note that labels MUST not collide when casing is ignored (see Case collision intolerance).

  13. suffix - An alphanumeric string that forms part of a filename, located after all entities and following a final _, right before the file extension; for example, it is eeg in sub-05_task-matchingpennies_eeg.vhdr.

  14. File extension - A portion of the filename after the left-most period (.) preceded by any other alphanumeric. For example, .gitignore does not have a file extension, but the file extension of test.nii.gz is .nii.gz. Note that the left-most period is included in the file extension.

  15. DEPRECATED - A "deprecated" entity or metadata field SHOULD NOT be used in the generation of new datasets. It remains in the standard in order to preserve the interpretability of existing datasets. Validating software SHOULD warn when deprecated practices are detected and provide a suggestion for updating the dataset to preserve the curator's intent.


An "entity" is an attribute that can be associated with a file, contributing to the identification of that file as a component of its filename in the form of a hyphen-separated key-value pair.

Each entity has the following attributes:

  1. Name: A comprehensive name describing the context of information to be provided via the entity.

  2. Key: A short string, typically a compression of the entity name, which uniquely identifies the entity when part of a filename.

  3. Value type: The requisite form of the value that gets specified alongside the key whenever the entity appears in a filename. For each entity, the value is of one of two possible types:

    1. Index: A non-negative integer, potentially zero-padded for consistent width.

    2. Label: An alphanumeric string. Note that labels MUST not collide when casing is ignored (see Case collision intolerance).

The entity format is a string that prescribes how the entity appears within any given filename. For a hypothetical entity with key "key", the format can be either "key-<index>" or "key-<label>", depending on the value type of that entity.

An entity instance is the specific manifestation of an entity within the name of a specific file, based on the format of the entity but with a value that provides identifying information to the particular file in whose name it appears.

Depending on context, any one of the entity name, key, format, or a specific entity instance, may be referred to as simply an "entity".

"Subject", "session", "sample", "task", and "run" from the list of definitions above are all examples of entities. The comprehensive list of supported entities is defined in the Entities Appendix; further, whether each is OPTIONAL, REQUIRED, or MUST NOT be provided for various data files, as well as their relative ordering in a filename, are defined in the Entity Tables Appendix.

Compulsory, optional, and additional data and metadata

The following standard describes a way of arranging data and writing down metadata for a subset of neuroimaging experiments. Some aspects of the standard are compulsory. For example a particular filename format is required when storing structural scans. Some aspects are regulated but optional. For example a T2 volume does not need to be included, but when it is available it should be saved under a particular filename specified in the standard. This standard aspires to describe a majority of datasets, but acknowledges that there will be cases that do not fit. In such cases one can include additional files and subdirectories to the existing directory structure following common sense. For example one may want to include eye tracking data in a vendor specific format that is not covered by this standard. The most sensible place to put it is next to the continuous recording file with the same naming scheme but different extensions. The solutions will change from case to case and publicly available datasets will be reviewed to include common data types in the future releases of the BIDS specification.

It is RECOMMENDED that non-compulsory metadata fields (like notch in channels.tsv files) and/or files (like events.tsv) are fully omitted when they are unavailable or unapplicable, instead of specified with an n/a value, or included as an empty file (for example an empty events.tsv file with only the headers included).

Dataset naming

BIDS does not prescribe a particular naming scheme for directories containing individual BIDS datasets. However, it is recommended to use a short descriptive name that reflects the content of the dataset, avoid spaces in the name, and use hyphens or underscores to separate words. BIDS datasets embedded within a larger BIDS dataset MAY follow some convention (see for example Storage of derived datasets).

Filesystem structure

Data for each subject are placed in subdirectories named "sub-<label>", where string "<label>" is substituted with the unique identification label of each subject. Additional information on each participant MAY be provided in a participants file in the root directory of the dataset.

If data for the subject were acquired across multiple sessions, then within the subject directory resides subdirectories named "ses-<label>", where string "<label>" is substituted with a unique identification label for each session. In datasets where at least one subject has more than one session, this additional subdirectory later SHOULD be added for all subjects in the dataset. Additional information on each session MAY be provided in a sessions file within the subject directory.

Within the session subdirectory (or the subject subdirectory if no session subdirectories are present) are subdirectories named according to data type as defined above. A data type directory SHOULD NOT be defined if there are no files to be placed in that directory.

Other top level directories

In addition to the subject directories, the root directory of a BIDS dataset MAY also contain the following directories:

  • code: A directory in which to store any code (for example the one used to generate the derivatives from the raw data). See the Code section for more information.
  • derivatives: Derivative data (for example preprocessed files). See the relevant section for more information.
  • phenotype: A directory where to store participant level measurements (for example responses from multiple questionnaires) split into individual files separate from participants.tsv. See the relevant section for more information.
  • sourcedata: A directory where to store data before harmonization, reconstruction, and/or file format conversion (for example, E-Prime event logs or DICOM files). See the relevant section for more information.
  • stimuli: A directory to store any stimulus files used during an experiment. See the relevant section for more information.


A filename consists of a chain of entity instances and a suffix all separated by underscores, and an extension. This pattern forms filenames that are both human- and machine-readable. For instance, file "sub-01_task-rest_eeg.edf" contains instances of the "subject" and "task" entities, making it evident from the filename alone that it contains resting-state data from subject 01; the suffix eeg and extension .edf depend on the imaging modality and the data format, and can therefore convey further details of the file's contents.

For a data file that was collected in a given session from a given subject, the filename MUST begin with the string sub-<label>_ses-<label>. Conversely, if the session level is omitted in the directory structure, the file name MUST begin with the string sub-<label>, without ses-<label>.

Any given entity MUST NOT appear more than once in any filename. For example, filename "sub-01_acq-laser_acq-uneven_electrodes.tsv" is invalid because it uses the "acquisition" entity twice.

In cases where an entity and a metadata field convey similar contextual information, the presence of an entity should not be used as a replacement for the corresponding metadata field. For instance, in echo-planar imaging MRI, the dir-<label> entity MAY be used to distinguish files with different phase-encoding directions, but the file's PhaseEncodingDirection MUST be specified as metadata.

A summary of all entities in BIDS and the order in which they MUST be specified is available in the entity table in the appendix.

Entity-linked file collections

An entity-linked file collection is a set of files that are related to each other based on a repetitive acquisition of sequential data by changing acquisition parameters one (or multiple) at a time or by being inherent components of the same data. Entity-linked collections are identified by a common suffix, indicating the group of files that should be considered a logical unit. Within each collection, files MUST be distinguished from each other by at least one entity (for example, echo) that corresponds to an altered acquisition parameter (EchoTime) or that defines a component relationship (for example, part). Note that these entities MUST be described by the specification and the parameter changes they declare MUST NOT invalidate the definition of the accompanying suffix. For example, the use of the echo entity along with the T1w suffix casts doubt on the validity of the identified contrast weighting. Provided the conditions above are satisfied, any suffix (such as bold) can identify an entity-linked file collection, although certain suffixes are exclusive for this purpose (for example, MP2RAGE). Use cases concerning this convention are compiled in the file collections appendix. This convention is mainly intended for but not limited to MRI modalities.

Case collision intolerance

Filename components are case sensitive, but collisions MUST be avoided when casing is ignored. For example, a dataset cannot contain both sub-s1 and sub-S1, as the labels would collide on a case-insensitive filesystem. Additionally, because the suffix eeg is defined, then the suffix EEG will not be added to future versions of the standard.

Uniqueness of data files

Data files MUST be uniquely identified by BIDS path components (entities, datatype, suffix). If multiple extensions are permissible (for example, .nii and .nii.gz), there MUST only be one such file with the same entities, datatype and suffix. This limitation does not apply to metadata files, such as JSON sidecar files or format-specific metadata files.

Note that duplicating files to make the same data available in multiple formats is not permitted. For example, if the files sub-01_ses-01_sample-A_photo.jpg and sub-01_ses-01_sample-A_photo.tif contain a representation of the same data, then the dataset MUST NOT contain both images. If the files contain different images, other entities MUST be used to distinguish the two.

Filesystem structure & Filenames richness versus distinctness

BIDS provides a rich filesystem structure and rich filenames by using entities, but it is important to keep in mind that files also have to be readable. They have to be readable by machines, and this implies that filenames cannot be longer than 255 characters. They also have to be readable by humans, and this implies minimizing length. A useful way to think about filenaming is distinctness: what is the minimal information needed to distinguish files? A simple illustration is given by using the ses- and run- entities. A T1 weighted MRI image could, in principle, be called sub-X_ses-1_run-1_T1w.nii. When there is only 1 session, and only 1 run, this is not needed as ses- and run- do not increase distinctiveness. Here, it is recommended to use the shorter version sub-X_T1w.nii. In some cases, this principle is enforced in the BIDS validator.

Source vs. raw vs. derived data

BIDS was originally designed to describe and apply consistent naming conventions to raw (unprocessed or minimally processed due to file format conversion) data. During analysis such data will be transformed and partial as well as final results will be saved. Derivatives of the raw data (other than products of DICOM to NIfTI conversion) MUST be kept separate from the raw data. This way one can protect the raw data from accidental changes by file permissions. In addition it is easy to distinguish partial results from the raw data and share the latter. See Storage of derived datasets for more on organizing derivatives.

Similar rules apply to source data, which is defined as data before harmonization, reconstruction, and/or file format conversion (for example, E-Prime event logs or DICOM files). Retaining the source data is especially valuable in a case when conversion fails to preserve crucial metadata unique to specific acquisition setup. Storing actual source files with the data is preferred over links to external source repositories to maximize long term preservation, which would suffer if an external repository would not be available anymore. This specification currently does not go into the details of recommending a particular naming scheme for including different types of source data (such as the raw event logs or parameter files, before conversion to BIDS). However, in the case that these data are to be included:

  1. These data MUST be kept in separate sourcedata directory. BIDS does not prescribe a particular naming scheme for source data, but it is recommended for it to follow BIDS naming convention where possible. For example: sourcedata/sub-01/ses-pre/func/sub-01_ses-pre_task-rest_bold.dicom.tgz or sourcedata/sub-01/ses-pre/func/MyEvent.sce.

  2. A README file SHOULD be found at the root of the sourcedata directory or the derivatives directory, or both. This file should describe the nature of the raw data or the derived data. We RECOMMEND including the PDF print-out with the actual sequence parameters generated by the scanner in the sourcedata directory.

Alternatively one can organize their data in the following way

└─ my_project-1/
   ├─ sourcedata/
   │  ├─ dicoms/
   │  ├─ raw/
   │  │  ├─ sub-01/
   │  │  ├─ sub-02/
   │  │  ├─ ... 
   │  │  └─ dataset_description.json 
   │  └─ ... 
   └─ derivatives/
      ├─ pipeline_1/
      ├─ pipeline_2/
      └─ ... 

In this example, sourcedata/dicoms is not nested inside sourcedata/raw, and only the sourcedata/raw subdirectory is a BIDS-compliant dataset among sourcedata/ subfolders. The subdirectories of derivatives MAY be BIDS-compliant derivatives datasets (see Non-compliant derivatives for further discussion). The above example is just a convention useful for organizing source, raw BIDS, and derived BIDS data while maintaining BIDS compliance of the raw data directory. When using this convention it is RECOMMENDED to set the SourceDatasets field in dataset_description.json of each subdirectory of derivatives to:

  "SourceDatasets": [ {"URL": "../../sourcedata/raw/"} ]


Sharing source data may help amend errors and missing data discovered only with the reuse of the raw dataset in practice. Therefore, from an Open Science perspective, it is RECOMMENDED to share the source data whenever it is possible.

However, more stringent sharing limitations may apply to the source data than those applicable to the raw data. For example, human data almost always requires deidentification before they can be redistributed, or the subjects' consent form did not explicitly state that the source files would be shared after deidentification. Further examples in which sharing source data may not be possible include original data formats that are not redistributable as per the acquisition device's license.

As for raw data, all regulatory, ethical, and legal aspects SHOULD be carefully considered before sharing data through the sourcedata/ directory mechanism. In the case of source data, these aspects are likely more stringent.

Storage of derived datasets

Derivatives can be stored/distributed in two ways:

  1. Under a derivatives/ subdirectory in the root of the source BIDS dataset directory to make a clear distinction between raw data and results of data processing. A data processing pipeline will typically have a dedicated directory under which it stores all of its outputs. Different components of a pipeline can, however, also be stored under different subdirectories. There are few restrictions on the directory names; it is RECOMMENDED to use the format <pipeline>-<variant> in cases where it is anticipated that the same pipeline will output more than one variant (for example, AFNI-blurring and AFNI-noblurring). For the sake of consistency, the subdirectory name SHOULD be the GeneratedBy.Name field in data_description.json, optionally followed by a hyphen and a suffix (see Derived dataset and pipeline description).

    Example of derivatives with one directory per pipeline:


    Example of a pipeline with split derivative directories:


    Example of a pipeline with nested derivative directories:

  2. As a standalone dataset independent of the source (raw or derived) BIDS dataset. This way of specifying derivatives is particularly useful when the source dataset is provided with read-only access, for publishing derivatives as independent bodies of work, or for describing derivatives that were created from more than one source dataset. The sourcedata/ subdirectory MAY be used to include the source dataset(s) that were used to generate the derivatives. Likewise, any code used to generate the derivatives from the source data MAY be included in the code/ subdirectory.

    Example of a derivative dataset including the raw dataset as source:

└─ my_processed_data/
   ├─ code/
   │  ├─ processing_pipeline-1.0.0.img 
   │  ├─ 
   │  └─ ... 
   ├─ sourcedata/
   │  ├─ sub-01/
   │  ├─ sub-02/
   │  └─ ... 
   ├─ sub-01/
   ├─ sub-02/
   ├─ ... 
   └─ dataset_description.json 

Throughout this specification, if a section applies particularly to derivatives, then Case 1 will be assumed for clarity in templates and examples, but removing /derivatives/<pipeline> from the template name will provide the equivalent for Case 2. In both cases, every derivatives dataset is considered a BIDS dataset and must include a dataset_description.json file at the root level (see Dataset description). Consequently, files should be organized to comply with BIDS to the full extent possible (that is, unless explicitly contradicted for derivatives). Any subject-specific derivatives should be housed within each subject's directory; if session-specific derivatives are generated, they should be deposited under a session subdirectory within the corresponding subject directory; and so on.

Non-compliant derivatives

Nothing in this specification should be interpreted to disallow the storage/distribution of non-compliant derivatives of BIDS datasets. In particular, if a BIDS dataset contains a derivatives/ subdirectory, the contents of that directory may be a heterogeneous mix of BIDS Derivatives datasets and non-compliant derivatives.

File format specification

Imaging files

All imaging data MUST be stored using the NIfTI file format. We RECOMMEND using compressed NIfTI files (.nii.gz), either version 1.0 or 2.0. If using compressed files, the gzip header SHOULD lack source filenames and timestamps. Imaging data SHOULD be converted to the NIfTI format using a tool that provides as much of the NIfTI header information (such as orientation and slice timing information) as possible. Since the NIfTI standard offers limited support for the various image acquisition parameters available in DICOM files, we RECOMMEND that users provide additional meta information extracted from DICOM files in a sidecar JSON file (with the same filename as the .nii[.gz] file, but with a .json extension). Extraction of BIDS compatible metadata can be performed using dcm2niix and dicm2nii DICOM to NIfTI converters. The BIDS-validator will check for conflicts between the JSON file and the data recorded in the NIfTI header.

Tabular files

Tabular data MUST be saved as plain-text, tab-delimited values (TSV) files (with extension .tsv), that is, CSV files where commas are replaced by tab characters. Tabs MUST be true tab characters and MUST NOT be a series of space characters. Tabular data such as continuous physiology recordings typically containing large numbers of rows MAY be saved as compressed tabular files (with extension .tsv.gz), which are introduced below. Plain-text TSV and compressed TSV are not interchangeable, that is, each section of the specification prescribes which one MUST be used for the data type at hand. Each TSV file MUST start with a header line listing the names of all columns with two exceptions:

  1. compressed tabular files, for which column names are defined in a sidecar metadata JSON object described below; and
  2. motion recording data, which use plain-text TSV and columns are defined as described in its corresponding section of the specifications.

It is RECOMMENDED that the column names in the header of the TSV file are written in snake_case with the first letter in lower case (for example, variable_name, not Variable_name). Column names defined in the header MUST be separated with tabs as for the data contents. Furthermore, column names MUST NOT be blank (that is, an empty string) and MUST NOT be duplicated within a single TSV file. String values containing tabs MUST be escaped using double quotes. Missing and non-applicable values MUST be coded as n/a. Numerical values MUST employ the dot (.) as decimal separator and MAY be specified in scientific notation, using e or E to separate the significand from the exponent. TSV files MUST be in UTF-8 encoding.


onset   duration    response_time   trial_type        trial_extra
200     20.0        15.8            word              中国人
240     5.0         17.34e-1        visual            n/a


The TSV examples in this document (like the one above this note) are occasionally formatted using space characters instead of tabs to improve human readability. Directly copying and then pasting these examples from the specification for use in new BIDS datasets can lead to errors and is discouraged.

Tabular files MAY be optionally accompanied by a simple data dictionary in the form of a JSON object within a JSON file. The JSON files containing the data dictionaries MUST have the same name as their corresponding tabular files but with .json extensions. If a data dictionary is provided, it MAY contain one or more fields describing the columns found in the TSV file (in addition to any other metadata one wishes to include that describe the file as a whole). Note that if a field name included in the data dictionary matches a column name in the TSV file, then that field MUST contain a description of the corresponding column, using an object containing the following fields:

Key name Requirement Level Data type Description
LongName OPTIONAL string Long (unabbreviated) name of the column.
Description RECOMMENDED string Free-form natural language description. The description of the column.
Levels RECOMMENDED object For categorical variables: An object of possible values (keys) and their descriptions (values).
Units RECOMMENDED string Measurement units for the associated file. SI units in CMIXF formatting are RECOMMENDED (see Units).
Delimiter OPTIONAL string If rows in a column may be interpreted as a lists of values, the character that separates one value from the next.
TermURL RECOMMENDED string URL pointing to a formal definition of this type of data in an ontology available on the web. For example: for "male".
HED OPTIONAL string or object of strings Hierarchical Event Descriptor (HED) information, see the HED Appendix for details.

Please note that while both Units and Levels are RECOMMENDED, typically only one of these two fields would be specified for describing a single TSV file column.


  "test": {
    "LongName": "Education level",
    "Description": "Education level, self-rated by participant",
    "Levels": {
      "1": "Finished primary school",
      "2": "Finished secondary school",
      "3": "Student at university",
      "4": "Has degree from university"
  "bmi": {
    "LongName": "Body mass index",
    "Units": "kg/m^2",
    "TermURL": ""

Each level can be described with a string as in the example above, or with an object containing the fields Description and TermURL like in the example below.

    "sex": {
        "Description": "sex of the participant as reported by the participant",
        "Levels": {
            "M": {
                "Description": "Male",
                "TermURL": ""
            "F": {
                "Description": "Female",
                "TermURL": ""

Compressed tabular files

Large tabular information, such as physiological recordings, MUST be stored with compressed tab-delineated (TSV.GZ) files when so established by the specifications. Rules for formatting plain-text tabular files apply to TSVGZ files with three exceptions:

  1. The contents of TSVGZ files MUST be compressed with gzip.
  2. Compressed tabular files MUST NOT contain a header in the first row indicating the column names.
  3. TSVGZ files MUST have an associated JSON file that defines the columns in the tabular file.


In contrast to plain-text TSV files, compressed tabular files files MUST NOT include a header line. Column names MUST be provided in the JSON file with the Columns field. Each column MAY additionally be described with a column description, as described in Tabular files.

TSVGZ are header-less to improve compatibility with existing software (for example, FSL, or PNM), and to facilitate the support for other file formats in the future.

Key-value files (dictionaries)

JavaScript Object Notation (JSON) files MUST be used for storing key-value pairs. JSON files MUST be in UTF-8 encoding. Extensive documentation of the format can be found at, and at Several editors have built-in support for JSON syntax highlighting that aids manual creation of such files. An online editor for JSON with built-in validation is available at It is RECOMMENDED that keys in a JSON file are written in CamelCase with the first letter in upper case (for example, SamplingFrequency, not samplingFrequency). Note however, when a JSON file is used as an accompanying sidecar file for a TSV file, the keys linking a TSV column with their description in the JSON file need to follow the exact formatting as in the TSV file.

Example of a hypothetical *_bold.json file, accompanying a *_bold.nii file:

  "RepetitionTime": 3,
  "Instruction": "Lie still and keep your eyes open"

Example of a hypothetical *_events.json file, accompanying an *_events.tsv file. Note that the JSON file contains a key describing an arbitrary column stim_presentation_side in the TSV file it accompanies. See task events section for more information.

  "stim_presentation_side": {
    "Levels": {
      "1": "stimulus presented on LEFT side",
      "2": "stimulus presented on RIGHT side"

The Inheritance Principle

In some circumstances, there can be multiple data files for which all or a subset of the relevant metadata is precisely equivalent. Where this occurs, it may be preferable to define those metadata only once, and be placed on the filesystem in such a way that those files are deemed to be applicable to each relevant data file individually, but not be erroneously associated with other data files to which the metadata contained within are not applicable. The Inheritance Principle defines a systematized set of rules to determine which metadata files to associate with which data files. Further, because multiple metadata files may apply to an individual data file, the Principle defines the order of precedence of such metadata files contents.


  1. Any metadata file (such as .json, .bvec or .tsv) MAY be defined at any directory level.

  2. For a given data file, any metadata file is applicable to that data file if:

    1. It is stored at the same directory level or higher;
    2. The metadata and the data filenames possess the same suffix;
    3. The metadata filename does not include any entity absent from the data filename.
  3. A metadata file MUST NOT have a filename that would be otherwise applicable to some data file based on rules 2.b and 2.c but is made inapplicable based on its location in the directory structure as per rule 2.a.

  4. There MUST NOT be multiple metadata files applicable to a data file at one level of the directory hierarchy.

  5. If multiple metadata files satisfy criteria 2.a-c above:

    1. For tabular files and other simple metadata files (for instance, bvec / bval files for diffusion MRI), accessing metadata associated with a data file MUST consider only the applicable file that is lowest in the filesystem hierarchy.

    2. For JSON files, key-values are loaded from files from the top of the directory hierarchy downwards, such that key-values from the top level are inherited by all data files at lower levels to which it is applicable unless overridden by a value for the same key present in another metadata file at a lower level (though it is RECOMMENDED to minimize the extent of such overrides).


  1. As per rule 3, metadata files applicable only to a specific participant / session MUST be defined in or below the directory corresponding to that participant / session; similarly, a metadata file that is applicable to multiple participants / sessions MUST NOT be placed within a directory corresponding to only one such participant / session.

  2. It is permissible for a single metadata file to be applicable to multiple data files at that level of the hierarchy or below. Where such metadata content is consistent across multiple data files, it is RECOMMENDED to store metadata in this way, rather than duplicating that metadata content across multiple metadata files.

  3. Where multiple applicable JSON files are loaded as per rule 5.b, key-values can only be overwritten by files lower in the filesystem hierarchy; the absence of a key-value in a later file does not imply the "unsetting" of that field (indeed removal of existing fields is not possible).


Example 1: Demonstration of inheritance principle

├─ sub-01/
│  └─ func/
│     ├─ sub-01_task-rest_acq-default_bold.nii.gz 
│     ├─ sub-01_task-rest_acq-longtr_bold.nii.gz 
│     └─ sub-01_task-rest_acq-longtr_bold.json 
└─ task-rest_bold.json 

Contents of file task-rest_bold.json:

    "EchoTime": 0.040,
    "RepetitionTime": 1.0

Contents of file sub-01/func/sub-01_task-rest_acq-longtr_bold.json:

    "RepetitionTime": 3.0

When reading image sub-01/func/sub-01_task-rest_acq-default_bold.nii.gz, only metadata file task-rest_bold.json is read; file sub-01/func/sub-01_task-rest_acq-longtr_bold.json is inapplicable as it contains entity "acq-longtr" that is absent from the image path (rule 2.c). When reading image sub-01/func/sub-01_task-rest_acq-longtr_bold.nii.gz, metadata file task-rest_bold.json at the top level is read first, followed by file sub-01/func/sub-01_task-rest_acq-longtr_bold.json at the bottom level (rule 5.b); the value for field "RepetitionTime" is therefore overridden to the value 3.0. The value for field "EchoTime" remains applicable to that image, and is not unset by its absence in the metadata file at the lower level (rule 5.b; corollary 3).

Example 2: Impermissible use of multiple metadata files at one directory level (rule 4)

└─ sub-01/
   └─ ses-test/
      ├─ anat/
      │  └─ sub-01_ses-test_T1w.nii.gz 
      └─ func/
         ├─ sub-01_ses-test_task-overtverbgeneration_run-1_bold.nii.gz 
         ├─ sub-01_ses-test_task-overtverbgeneration_run-2_bold.nii.gz 
         ├─ sub-01_ses-test_task-overtverbgeneration_bold.json 
         └─ sub-01_ses-test_task-overtverbgeneration_run-2_bold.json 

Example 3: Modification of filesystem structure from Example 2 to satisfy inheritance principle requirements

└─ sub-01/
   └─ ses-test/
      ├─ sub-01_ses-test_task-overtverbgeneration_bold.json 
      ├─ anat/
      │  └─ sub-01_ses-test_T1w.nii.gz 
      └─ func/
         ├─ sub-01_ses-test_task-overtverbgeneration_run-1_bold.nii.gz 
         ├─ sub-01_ses-test_task-overtverbgeneration_run-2_bold.nii.gz 
         └─ sub-01_ses-test_task-overtverbgeneration_run-2_bold.json 

Example 4: Single metadata file applying to multiple data files (corollary 2)

└─ sub-01/
   ├─ anat/
   └─ func/
      ├─ sub-01_task-xyz_acq-test1_run-1_bold.nii.gz 
      ├─ sub-01_task-xyz_acq-test1_run-2_bold.nii.gz 
      └─ sub-01_task-xyz_acq-test1_bold.json 

Participant names and other labels

BIDS allows for custom user-defined <label>s and <index>es for example, for naming of participants, sessions, acquisition schemes. Note that they MUST consist only of allowed characters as described in Definitions above. In <index>es we RECOMMEND using zero padding (for example, 01 instead of 1 if some participants have two-digit labels) to make alphabetical sorting more intuitive. Note that zero padding SHOULD NOT be used to merely maintain uniqueness of <index>es.

Please note that a given label or index is distinct from the "prefix" it refers to. For example sub-01 refers to the sub entity (a subject) with the label 01. The sub- prefix is not part of the subject label, but must be included in filenames (similarly to other entities).

Specification of paths

Several metadata fields in BIDS require the specification of paths, that is, a string of characters used to uniquely identify a location in a directory structure. For example the IntendedFor or AssociatedEmptyroom metadata fields.

Throughout BIDS all such paths MUST be specified using the slash character (/), regardless of the operating system that a particular dataset is curated on or used on.

Paths SHOULD NOT be absolute local paths, because these might break when a dataset is used on a different machine. It is RECOMMENDED that all paths specified in a BIDS dataset are relative paths, as specified in the respective descriptions of metadata fields that require the use of paths.

Uniform Resource Indicator

A Uniform Resource Indicator (URI) is a string referring to a resource and SHOULD have the form <scheme>:[//<authority>]<path>[?<query>][#<fragment>], as specified in RFC 3986. This applies to URLs and other common URIs, including Digital Object Identifiers (DOIs), which may be fully specified as doi:<path>, for example, doi:10.5281/zenodo.3686061. A given resource may have multiple URIs. When selecting URIs to add to dataset metadata, it is important to consider specificity and persistence.

Several fields are designated for DOIs, for example, DatasetDOI in dataset_description.json. DOI values SHOULD be fully specified URIs such as doi:10.18112/openneuro.ds000001.v1.0.0. Bare DOIs such as 10.18112/openneuro.ds000001.v1.0.0 are DEPRECATED.


To reference files in BIDS datasets, the following URI scheme may be used:


The scheme component bids identifies a BIDS URI, which defines a path component of the form <dataset-name>:<relative-path>. The dataset-name component is an identifier for a BIDS dataset, and the relative-path component is the location of a resource within that BIDS dataset, relative to the root of that dataset. The relative-path MUST NOT start with a forward-slash character (/).



If no dataset name is specified, the URI is relative to the current BIDS dataset. This is made more precise in the next section.

Resolution of BIDS URIs

In order to resolve a BIDS URI, the dataset name must be mapped to a BIDS dataset.

The special case "" (that is, the empty string) refers to the BIDS dataset in which the BIDS URI is found. The dataset root is the nearest parent directory that contains a valid dataset_description.json.

All other dataset names MUST be specified in the DatasetLinks object in dataset_description.json, which maps dataset names to URIs that point to BIDS dataset locations. If the scheme is omitted from a URI in DatasetLinks, that path is resolved relative to the current dataset root (see deriv1 example, below).

BIDS URIs cannot be interpreted outside a BIDS dataset, as they require a dataset_description.json file to resolve.


Consider this example dataset_description.json:

    "DatasetLinks": {
        "deriv1": "derivatives/derivative1",
        "phantoms": "file:///data/phantoms",
        "ds000001": "doi:10.18112/openneuro.ds000001.v1.0.0"

Here deriv1 refers to a BIDS Derivatives dataset contained within the current dataset, phantoms refers to a BIDS dataset of phantom data stored on the local filesystem, and ds000001 refers to a BIDS dataset that must be resolved by DOI.

Note that resolving bids:phantoms:sub-phantom01/anat/sub-phantom01_T1w.nii.gz is a straightforward concatenation: file:///data/phantoms/sub-phantom01/anat/sub-phantom01_T1w.nii.gz. However, retrieving bids:ds000001:sub-02/anat/sub-02_T1w.nii.gz requires first resolving the DOI, identifying the retrieval method, possibly retrieving the entire dataset, and finally constructing a URI to the desired resource.

No protocol is currently proposed to automatically resolve all possible BIDS URIs.

Future statement

BIDS URIs are parsable as standard URIs with scheme bids and path [<dataset-name>]:<relative-path>. The authority, query and fragment components are unused. Future versions of BIDS may specify interpretations for these components, but MUST NOT change the interpretation of a previously valid BIDS URI. For example, a future version may specify an authority that would allow BIDS URIs to be resolved without reference to a local dataset_description.json.


All units SHOULD be specified as per International System of Units (abbreviated as SI, from the French Système international (d'unités)) and can be SI units or SI derived units. In case there are valid reasons to deviate from SI units or SI derived units, the units MUST be specified in the sidecar JSON file. In case data is expressed in SI units or SI derived units, the units MAY be specified in the sidecar JSON file. In case non-standard prefixes are added to SI or non-SI units, these non-standard prefixed units MUST be specified in the JSON file. See the Units Appendix for a list of standard units and prefixes. Note also that for the formatting of SI units, the CMIXF-12 convention for encoding units is RECOMMENDED. CMIXF provides a consistent system for all units and prefix symbols with only basic characters, avoiding symbols that can cause text encoding problems; for example the CMIXF formatting for "micro volts" is uV, "degrees Celsius" is oC and "Ohm" is Ohm. See the Units Appendix for more information.

For additional rules, see below:

  • Elapsed time SHOULD be expressed in seconds. Please note that some DICOM parameters have been traditionally expressed in milliseconds. Those need to be converted to seconds.

  • Frequency SHOULD be expressed in Hertz.

  • Arbitrary units SHOULD be indicated with the string "arbitrary".

Describing dates and timestamps:

  • Date time information MUST be expressed in the following format YYYY-MM-DDThh:mm:ss[.000000][Z] (year, month, day, hour (24h), minute, second, optional fractional seconds, and optional UTC time indicator). This is almost equivalent to the RFC3339 "date-time" format, with the exception that UTC indicator Z is optional and non-zero UTC offsets are not indicated. If Z is not indicated, time zone is always assumed to be the local time of the dataset viewer. No specific precision is required for fractional seconds, but the precision SHOULD be consistent across the dataset. For example 2009-06-15T13:45:30.

  • Time stamp information MUST be expressed in the following format: hh:mm:ss[.000000] For example 13:45:30.

  • Note that, depending on local ethics board policy, date time information may not need to be fully detailed. For example, it is permissible to set the time to 00:00:00 if reporting the exact recording time is undesirable. However, for privacy protection reasons, it is RECOMMENDED to shift dates, as described below, without completely removing time information, as time information can be useful for research purposes.

  • Dates can be shifted by a random number of days for privacy protection reasons. To distinguish real dates from shifted dates, is is RECOMMENDED to set shifted dates to the year 1925 or earlier. Note that some data formats do not support arbitrary recording dates. For example, the EDF data format can only contain recording dates after 1985. For longitudinal studies dates MUST be shifted by the same number of days within each subject to maintain the interval information. For example: 1867-06-15T13:45:30

  • WARNING: The Neuromag/Elekta/MEGIN file format for MEG (.fif) does not support recording dates earlier than 1902 roughly. Some analysis software packages (for example, MNE-Python) handle their data as .fif internally and will break if recording dates are specified prior to 1902, even if the original data format is not .fif. See the MEG File Formats Appendix for more information.

  • Age SHOULD be given as the number of years since birth at the time of scanning (or first scan in case of multi session datasets). Using higher accuracy (weeks) should in general be avoided due to privacy protection, unless when appropriate given the study goals, for example, when scanning babies.

Directory structure

Single session example

This is an example of the directory and file structure. Because there is only one session, the session level is not required by the format. For details on individual files see descriptions in the next section:

├─ sub-control01/
│  ├─ anat/
│  │  ├─ sub-control01_T1w.nii.gz 
│  │  ├─ sub-control01_T1w.json 
│  │  ├─ sub-control01_T2w.nii.gz 
│  │  └─ sub-control01_T2w.json 
│  ├─ func/
│  │  ├─ sub-control01_task-nback_bold.nii.gz 
│  │  ├─ sub-control01_task-nback_bold.json 
│  │  ├─ sub-control01_task-nback_events.tsv 
│  │  ├─ sub-control01_task-nback_physio.tsv.gz 
│  │  ├─ sub-control01_task-nback_physio.json 
│  │  └─ sub-control01_task-nback_sbref.nii.gz 
│  ├─ dwi/
│  │  ├─ sub-control01_dwi.nii.gz 
│  │  ├─ sub-control01_dwi.bval 
│  │  └─ sub-control01_dwi.bvec 
│  └─ fmap/
│     ├─ sub-control01_phasediff.nii.gz 
│     ├─ sub-control01_phasediff.json 
│     └─ sub-control01_magnitude1.nii.gz 
├─ code/
│  └─ 
├─ derivatives/
├─ participants.tsv 
├─ dataset_description.json 

Unspecified data

Additional files and directories containing raw data MAY be added as needed for special cases. All non-standard file entities SHOULD conform to BIDS-style naming conventions, including alphabetic entities and suffixes and alphanumeric labels/indices. Non-standard suffixes SHOULD reflect the nature of the data, and existing entities SHOULD be used when appropriate. For example, an ASSET calibration scan might be named sub-01_acq-ASSET_calibration.nii.gz.

Non-standard files and directories should be named with care. Future BIDS efforts may standardize new entities and suffixes, changing the meaning of filenames and setting requirements on their contents or metadata. Validation and parsing tools MAY treat the presence of non-standard files and directories as an error, so consult the details of these tools for mechanisms to suppress warnings or provide interpretations of your filenames.