OpenNeuro is a free, open-access data-sharing platform for human neuroimaging research, built specifically around the Brain Imaging Data Structure (BIDS) standard. Researchers use it to deposit raw and preprocessed brain-imaging datasets for public reuse, and journals and funders increasingly point to it as an acceptable repository for satisfying open-data requirements in neuroscience. This guide covers what OpenNeuro is, how it relates to BIDS, what kinds of data it accepts, and where it sits in the wider neuroimaging data-sharing landscape.
What OpenNeuro is
OpenNeuro is hosted and maintained by the Stanford Center for Reproducible Neuroscience, led by Russell Poldrack, and runs on Amazon Web Services (AWS) — it has been part of the AWS Public Datasets program, which provides free storage and bandwidth for openly available scientific datasets. The platform launched publicly in June 2017 as the successor to an earlier Stanford-run project, OpenfMRI. The rename signaled a deliberate scope expansion: where OpenfMRI accepted only functional MRI data, OpenNeuro was built to accept any imaging modality that can be expressed in the BIDS format, not fMRI alone. Development was carried out by the Center working with Squishymedia, with funding support from the National Science Foundation, the National Institute on Drug Abuse, and the Laura and John Arnold Foundation.
The platform’s stated purpose is to improve the transparency and reproducibility of neuroscience research by making datasets openly and persistently available for reuse, reanalysis, and methods validation — rather than leaving raw imaging data to live only on a lab’s local servers until a paper is published, if it is shared at all.
How OpenNeuro relates to BIDS
OpenNeuro is built around, and enforces, the BIDS standard rather than treating it as one optional format among several. The BIDS specification is a community-developed convention, originally designed for MRI data and since extended to other modalities, that specifies how a neuroimaging dataset’s files and folders should be named and organized and how the metadata describing each scan should be recorded. Concretely, BIDS requires a predictable nested directory structure (organized by subject, and within each subject by session and imaging modality), standardized filenames built from a defined set of key-value pairs, and a JSON “sidecar” file alongside each image file that documents scan parameters and other metadata a downstream user needs to interpret and reprocess the data correctly.
Standardizing this organization matters for reproducibility because inconsistent, lab-specific data layouts are themselves a barrier to reuse: a script or analysis pipeline written for one lab’s folder structure typically cannot run against another lab’s dataset without manual reformatting, which costs time and introduces opportunities for error. A shared structure means the same analysis software, including the community-maintained BIDS Apps ecosystem of containerized processing pipelines, can run against any BIDS-compliant dataset without dataset-specific adaptation. OpenNeuro operationalizes this by requiring every uploaded dataset to pass the BIDS validator before it is accepted — a dataset that isn’t BIDS-compliant cannot be published on the platform. This validation-at-deposit approach is one of the more concrete examples in research data management of a repository enforcing a FAIR-relevant metadata standard as a hard gate rather than a recommendation.
What data types OpenNeuro hosts
OpenNeuro accepts BIDS-compliant datasets spanning several human neuroimaging modalities:
- MRI — including structural (anatomical) MRI and functional MRI (fMRI)
- PET (positron emission tomography)
- MEG (magnetoencephalography)
- EEG (electroencephalography)
- iEEG (intracranial electroencephalography)
This breadth reflects BIDS’s own extension history: the specification was originally written for MRI, and separate BIDS extensions were subsequently developed to cover MEG, EEG, and iEEG (and, more recently, other modalities such as near-infrared spectroscopy and microscopy data, though those are less commonly represented on OpenNeuro specifically). Because every dataset on the platform follows the same underlying organizational logic regardless of modality, cross-modal or multimodal studies (for example, a dataset combining structural MRI with EEG from the same participants) can still be represented consistently within a single BIDS dataset.
Open-access model
Datasets uploaded to OpenNeuro are made available to the public with minimal restrictions, intended to maximize reuse rather than gate access behind data use agreements or approval workflows. This open-by-default posture is what allows OpenNeuro to function not just as a private lab archive but as a genuine community resource: other researchers can download a dataset, rerun the original analysis to check computational reproducibility, or reuse the raw data for an entirely different secondary analysis or methods paper. Because BIDS datasets carry rich, standardized metadata alongside the imaging data itself, a dataset downloaded from OpenNeuro is more likely to be interpretable and reusable by someone outside the originating lab than an equivalent dataset shared in an ad hoc format — the open-access model and the BIDS requirement reinforce each other.
Researchers depositing data containing human subjects still need to have handled consent and de-identification appropriately before upload; OpenNeuro’s open posture governs what happens to a dataset once it has been prepared for sharing, not a substitute for an institution’s own human-subjects and data-sharing review.
Where OpenNeuro fits in the neuroimaging data-sharing ecosystem
OpenNeuro is a domain-specific (disciplinary) repository rather than a generalist one: it exists specifically for BIDS-formatted human neuroimaging data, in contrast to generalist repositories such as Figshare or Dryad, which accept research data from any discipline in whatever format the depositor provides. That specialization is precisely why it can enforce a single, strict data standard at intake — a generalist repository has no comparable basis for requiring every deposit to follow one field-specific structure. For a research administrator helping a neuroimaging lab choose where to deposit data, that means OpenNeuro is typically the more defensible choice whenever a funder or journal open-data mandate can be satisfied by a discipline-appropriate repository, since it gives reviewers and reusers a standardized, validated dataset rather than an unstructured file dump. See CASRAI’s guide on how to choose an open data repository for the general framework behind that kind of decision.
OpenNeuro also sits inside a broader BIDS-centered tooling ecosystem: BIDS Apps provide standardized, containerized analysis pipelines that run directly against BIDS datasets, and other infrastructure (such as DataLad-based version control) has been built to work with BIDS-organized data specifically because the format is predictable. Within the NIH BRAIN Initiative’s data ecosystem, OpenNeuro has been supported as an open archive for sharing BRAIN Initiative-funded imaging data, reflecting its role as recognized infrastructure rather than a one-off lab tool. For research administrators tracking reproducibility commitments more broadly, see CASRAI’s overview of reproducibility infrastructure, workflows, and containers, and the pillar page on research data management for how repository selection fits into the wider RDM lifecycle.
Frequently asked questions
Does OpenNeuro charge to deposit or download data?
No. OpenNeuro is a free platform for both depositing and downloading datasets; its cloud storage and bandwidth have been supported through the AWS Public Datasets program rather than depositor fees.
Can I deposit non-BIDS-formatted neuroimaging data on OpenNeuro?
No. Every dataset submitted to OpenNeuro is run through the BIDS validator and must be BIDS-compliant to be published on the platform. Researchers with data in another format need to convert it to BIDS first; a range of community conversion tools exist for this purpose.
Is OpenNeuro the same thing as BIDS?
No. BIDS is a data organization and metadata specification — a set of rules for how files should be named, structured, and documented. OpenNeuro is a specific hosting platform that requires uploaded datasets to follow that specification. Other repositories and local lab archives can also use BIDS without being OpenNeuro, and BIDS itself is maintained independently of any single repository.
What happened to OpenfMRI?
OpenfMRI was OpenNeuro’s predecessor, run by the same Stanford team and limited to fMRI data. It was retired in favor of OpenNeuro in 2017, when the platform was rebuilt around the broader, modality-agnostic BIDS standard.
Does using OpenNeuro satisfy a funder’s data-sharing or public-access mandate?
Depositing BIDS-compliant neuroimaging data on OpenNeuro is widely treated as satisfying open-data-sharing expectations from funders and journals in neuroscience, since it is a recognized, community-standard, openly accessible repository. Researchers should still check the specific wording of their funder’s or journal’s data-sharing policy — for example, NIH’s Data Management and Sharing Policy — to confirm a domain repository is an acceptable choice for their award, rather than assuming any repository automatically qualifies.
