Free, licensed and dated, because a number you cannot download is a number you cannot check.
938 registered devices and 385 removals, retrieved 2026-08-31, as CSV or JSON. This is the same data every figure on this site is computed from, published so you can reproduce any of them or disagree with us using the same rows.
Every device currently self-certified on the FMCSA registered list, with model, software version, ELD identifier, vendor and supported data transfer methods.
Every device FMCSA has published as removed, with the removal date and whether the vendor withdrew it or FMCSA revoked it.
The same registered-device data as structured JSON, with dataset metadata and the retrieval date.
Removals plus the diff against our previous snapshot, which is the part FMCSA does not publish.
CSV files are RFC 4180 with CRLF line endings and a header row. Any cell beginning with an equals sign, plus, minus or at sign is prefixed with an apostrophe so a spreadsheet treats it as text rather than a formula.
CC BY 4.0. Use it commercially, republish it, build on it. Attribute FreightProbe and link back.
Free, with no sign-up, no key and no rate limit. If you need it programmatically, the JSON files are the same URLs.
The source carries vendor email, phone and postal addresses. We deliberately exclude them. Use the FMCSA list itself to contact a vendor.
FreightProbe. "FMCSA Registered and Removed ELD Devices." Retrieved 2026-08-31. https://freightprobe.com/data
Quote the retrieval date with any figure. This dataset changes as FMCSA publishes registrations and removals, so a number without a date is not reproducible. We hold dated snapshots from 2026-08-31 onward.
Both files are parsed from the FMCSA registered and removed lists. Columns are matched by name rather than position, because the two lists have different column counts and a positional index that is correct on one is silently wrong on the other.
Two limits worth knowing before you compute anything. FMCSA publishes some companies under more than one name string, so grouping by company is approximate unless you normalise; our own pages disagree by three on the single-product vendor count for exactly this reason, and we say so rather than picking a number. And the published lists contain a small number of exactly duplicated rows, 8 in the registered file and 18 in the removals file, which we pass through unchanged rather than silently deduplicating.
Findings computed from this data: brands that outlive their devices, companies that left the registry, and shared hardware behind separate brands.