Accessing the data: what’s where

Accessing the data: what’s where#

Last reviewed: October 2026 (ABCD Release 7.0). Official sources: ABCD docs: Access, Data structure, NBDC resource hub.

Once your DUC is approved, ABCD data can be reached through a few different platforms. This page explains what each one is and what data it holds. For the mechanics of actually getting files onto your machine, see Downloading the data.

Releases#

Release

Released

Latest visit included

DOI

Where

7.0 (current)

May 2026

ses-07M (7.5-year)

10.82525/8f3w-5260

NBDC

6.1

Dec 2025

ses-06A

10.82525/ab4q-qr87

NBDC (Previous Releases)

6.0

Jun 2025

ses-06A

10.82525/jy7n-g441

NBDC (Previous Releases)

5.1 and earlier

Nov 2023

ses-04A

5.1.1: 10.82525/wjkw-bp63

NDA (legacy DUCs only)

Sources: release notes index, NBDC ABCD page.

Release 7.0 includes 11,860 participants and adds, among other things, whole-genome sequencing data and additional linked external data. Read the 7.0 release notes and known issues before you start. For example, T1w/T2w/fieldmap scans are missing for 889 Siemens XA30 sessions in 7.0.

Tip

Pick a release and stick with it for a project. Write the release number and DOI into your project README on day one. Previous releases stay available on NBDC, so you can always reproduce your work.

The platforms#

Platform

What it is

Data available

Needs DUC?

DEAP (docs)

Data Exploration & Analysis Portal: data dictionary, dataset builder, interactive plots, in-browser mixed models (FEMA)

Tabulated data (incl. tabulated imaging summaries)

Dictionary: no. Data: yes

NBDC Data Access Platform (Lasso)

Query tools plus the file-based data browser

Tabulated and file-based data (MRI, EEG, wearables, genetics, EARS, Fitbit), previous releases

Yes

Globus

Transfer service used for all large/bulk downloads

Everything on the Data Access Platform

Yes

NBDC Sandbox (Lasso SAFE)

Cloud compute without downloading: RStudio, Jupyter, Linux desktop, Slurm

Tabulated data mounted read-only

Yes

NBDC Biobank Portal

Biospecimen requests

Biospecimens

Separate request

Note

DEAP has two hostnames: abcd.deapscience.com (ABCD) and nbdc.deapscience.com (combined NBDC). Genetics, EARS (smartphone sensing), and Fitbit data are not in DEAP. Get them from the Data Access Platform/Globus.

Should I download at all, or use the Sandbox?#

The DUC encourages keeping data on the platform rather than downloading. The Sandbox is a good option if:

  • your institution doesn’t (yet) have a NIST 800-171 compliant environment,

  • you only need tabulated data, or

  • you want to avoid managing local copies and deleting them at the end.

At last review, Sandbox VMs ranged up to 16 cores / 128 GB RAM, with a 30 GB home directory. Tabulated data live at /shared/release/abcd/dairc_files/rawdata/phenotype as TSV and Parquet. Check which release the Sandbox currently mounts.

How the data are organized (6.0+)#

Release 6.0 reorganized ABCD around the BIDS standard, and 7.0 keeps that structure. If you learned ABCD on 5.1 or earlier, almost every name has changed.

File tree#

abcd/
├── rawdata/
│   ├── phenotype/            # tabulated data: <table>.tsv / .parquet + .json sidecar
│   └── sub-*/ses-*/          # NIfTI imaging, neurocognitive, wearable files
├── sourcedata/sub-*/ses-*/   # DICOMs (per-series .tgz)
├── derivatives/
│   ├── freesurfer/           # sub-<p>_ses-<e>.zip
│   ├── mmps_mproc/           # minimally processed images (DAIRC pipeline)
│   └── abcc-<pipeline>/      # ABCD-BIDS Community Collection derivatives
└── concatenated/imaging/     # correlation matrices, vertexwise & voxelwise data

(Data structure docs)

IDs and sessions#

  • participant_id: sub- + 8 characters (e.g., sub-ABCD1234)

  • session_id: ses- + 3 characters

New session_id

Old eventname (≤ 5.1)

ses-00S

screener_arm_1

ses-00A

baseline_year_1_arm_1

ses-00M

6_month_follow_up_arm_1

ses-01A

1_year_follow_up_y_arm_1

ses-02A … ses-06A

annual follow-ups (ses-0xM = mid-year visits)

ses-07M

7.5-year visit (new in 7.0)

ses-C01–ses-C07

COVID-19 survey waves

ses-S01–ses-S03

Social Development (SDev) substudy waves

See Standards for the complete list.

Tables and variables#

  • Static tables have one row per participant. Longitudinal tables have one row per participant × session (long format).

  • Variable names follow domain_source_table_item, e.g. ph_y_bp__dia__r01_001. Common suffixes: __v01 (changed version), __l (longitudinal version), ___1 (multi-select option). See Naming conventions.

  • Old ↔ new names: the data dictionary lists the old (≤ 5.1) name for each variable, and NBDCtools::convert_names_data() / convert_names_file() convert between them. This is handy for porting old scripts.

  • Missing-value codes: 222, 444, 555, 666, 777, 888, 999 are codes, not numbers. Convert them before analysis (e.g., NBDCtools::transf_value_to_na()).

  • Dates/times: variables ending in _dt are dates; _dtt are date-times.

  • Metadata: each table comes with a data dictionary (one row per variable) and a levels table (value labels), available in DEAP, on NBDC, and as BIDS JSON sidecars (metadata docs).

Imaging#

  • Raw BIDS files look like sub-<label>_ses-<label>_task-<mid|nback|sst|rest>_run-<label>_bold.nii.gz. scans.tsv carries usability/QC flags (raw data docs).

  • DAIRC tabulated imaging (ROI summaries) are in the phenotype tables, so you can get them via DEAP.

  • ABCC (ABCD-BIDS Community Collection) is now part of the official release, distributed via NBDC (v4.0.0: fMRIPrep, XCP-D, QSIPrep/QSIRecon). ABCD-HCP pipeline derivatives are no longer produced as of 7.0; legacy outputs remain in 6.1. ABCC 1.0–2.0 lived in NDA Collection 3165, and those docs are archived. See the ABCC start page.

Contents