Anyone can pull a carrier's MCS-150 record. The census file is a public download, SAFER renders the same fields on demand, and a dozen marketplace scrapers will hand you the raw payload for pennies per lookup. Possessing the data has been a commodity for years.
Understanding it has not.
The MCS-150 is one of the most misread datasets in freight. It looks like an operational snapshot. It reads like ground truth. It is neither. It is a self-reported regulatory filing, captured at irregular intervals, stored in a system designed for safety enforcement rather than commercial decisioning, and surfaced through layers that strip away every cue you would need to know how much to trust it. The fields are simple. The interpretation is not.
This guide walks the record field by field, then covers the four interpretation problems that raw data cannot solve on its own, and closes with what the 2026 Motus transition changes about all of it.
What the MCS-150 actually is, and is not
The MCS-150, officially the Motor Carrier Identification Report, is the filing that establishes and maintains a carrier's USDOT number record inside the Motor Carrier Management Information System (MCMIS). Before December 12, 2015, carriers used it to obtain a USDOT number. After that date, first-time applicants moved to the Unified Registration System, and the MCS-150 became primarily a maintenance instrument: the biennial update, plus a notice mechanism for changes to address, ownership, fleet size, or operations.
Two things follow from this that most consumers of the data miss.
First, it is self-reported. Every operational field on the form, power units, mileage, driver counts, cargo types, is a claim the carrier made about itself, not a measurement anyone verified. FMCSA captures what the carrier entered. It does not audit it at filing time.
Second, it is captured on a slow, staggered cadence. The biennial update is required every 24 months, and the schedule is keyed to the USDOT number itself: if the next-to-last digit is odd, the update is due in odd-numbered years; if even, in even-numbered years. The last digit sets the month. A carrier can voluntarily update more often, and many do when something material changes, but the floor is once every two years. That means at any given moment a meaningful share of "current" records describe a business as it existed up to 24 months ago. For a fast-moving small carrier, that is several lifetimes.
Note also that brokers and freight forwarders are exempt from the biennial update requirement even though they hold USDOT numbers and appear in the same census. Treating their records as if they followed the same refresh discipline as carriers is a common and avoidable error.
The chain of custody: form to census to scraper
To interpret a field, you have to know what happened to it on the way to you.
A carrier files the MCS-150 (historically on paper or through the FMCSA Portal, now increasingly through Motus). FMCSA keys it into MCMIS. MCMIS is the system of record. From there the same underlying data is exposed through several surfaces: the SAFER Company Snapshot, the downloadable MCMIS census file documented by the Volpe Center, and the Socrata open-data endpoints on data.transportation.gov.
Marketplace scrapers, the Apify actors and their equivalents, sit at the very end of this chain. They hit SAFER or the Socrata API at run time and hand back whatever those surfaces render, often packaged with a lead-generation pitch and email enrichment bolted on. That is a legitimate plumbing layer, and for some jobs it is all you need. But it is important to be clear about what it delivers: the raw field, exactly as FMCSA stores it, with no signal about that field's age, reliability, or relationship to the rest of the record. The scraper's job ends where interpretation begins.
The fields that matter, and how to read them
The census record has dozens of elements. Here are the ones that carry the most decisioning weight, and the traps in each.
Operating status (A / I)
The status flag marks a record Active or Inactive. Active means the entity is currently in business and subject to the safety regulations. Inactive means it is no longer in business or no longer subject to them.
The trap: operating status is not operating authority, and neither is insurance. A carrier can show Active census status while its for-hire authority is revoked or its insurance has lapsed. These live in different files (Licensing and Insurance, not the census) and answer different questions. "Active" in the census tells you the record is live, not that the carrier can legally haul your freight today.
Entity operation and classification (C / S / B / R / F / T, then A through L)
The operation field identifies what the entity is: Carrier, Shipper, Broker, Registrant, Freight Forwarder, or cargo Tank facility. An entity can carry more than one letter. A second field, classification, distinguishes Authorized For Hire (A), Exempt For Hire (B), Private property (C), and several government and passenger categories.
The trap: "Registrant" (R) is not a carrier. It is an entity that registers vehicles but does not operate as a carrier. Lead lists that filter on a USDOT number alone, without reading operation and classification, routinely pull registrants, private fleets, and shippers into a carrier prospecting motion. The letters are the difference between a real for-hire carrier and a private fleet that will never buy what you sell.
Power units (TOT_PWR, and the OWN / TRM / TRP split)
Fleet size is the field everyone wants and the one most often misused. The census stores total power units (TOT_PWR), total trucks, and total buses, built up from granular equipment counts. Those counts are themselves split three ways: owned (OWN), term-leased (TRM), and trip-leased (TRP). In the SAFER snapshot, the headline power-unit figure is also bucketed into letter ranges rather than reported as an exact number for some views.
The traps are layered. The number is self-reported and only as fresh as the last filing, so a carrier that has doubled or folded since its last update will show its old size. The owned-versus-leased split materially changes what the number means: a "20 power unit" carrier that trip-leases most of its capacity is a different commercial animal from one that owns its trucks. And because the figure feeds safety exposure calculations, carriers have incentives that pull the number in non-obvious directions. Reading TOT_PWR without reading the OWN/TRM/TRP composition and the filing date is reading one number out of three.
Drivers (interstate / intrastate, within and beyond 100 miles, CDL)
The census breaks drivers into interstate and intrastate, each split by within or beyond a 100-mile radius, plus a CDL count and an average monthly trip-leased driver figure.
The trap: the radius split is a genuine operational signal that almost no consumer uses. A carrier whose drivers are overwhelmingly intrastate within 100 miles runs a fundamentally different operation, regional or local, than one weighted to interstate beyond 100 miles. Collapsing this to a single driver count throws away the most useful thing the field contains.
Mileage (MCS150_MILEAGE and MCS150MILEAGEYEAR)
This is the single most abused field in the entire record, and the one that separates people who understand the data from people who merely have it.
The mileage figure is vehicle miles traveled, self-reported by the carrier. Critically, it is stamped with a separate mileage-year field recording which year the miles describe. The filing date and the mileage year are not the same thing, and neither is "now." A 2026 filing can report 2023 mileage. Many records carry mileage that is years stale or blank entirely.
This matters beyond curiosity because mileage drives safety exposure. In FMCSA's Safety Measurement System, when reported VMT is too old it is dropped from the calculation, and exposure defaults to an average of power units over a trailing window instead. So the same number you might use as a proxy for "how active is this carrier" is, inside FMCSA's own systems, treated as untrustworthy past a certain age. Using raw MCS150_MILEAGE without reading MCS150MILEAGEYEAR alongside it produces confidently wrong activity estimates. The two fields are meaningless apart.
Cargo classifications
The form captures up to 28 declared cargo types, from General Freight and Household Goods through Refrigerated Food, Liquids and Gases, and Livestock.
The trap: these are declared categories, not observed hauling behavior. Carriers commonly check many boxes at registration to preserve flexibility and never revisit them. The cargo array tells you what a carrier said it might haul, not what it actually moves. Lane and commodity inference from this field alone is guesswork dressed as data.
Addresses, phone, and company representatives
The record carries a physical address (the principal place of business) and a separate mailing address, which may be a PO box, plus phone, cell, and fax, and one or two named company representatives. Federal tax identifiers (EIN and SSN) exist in MCMIS but are withheld from public dissemination for privacy.
The trap, and the opportunity: the physical-versus-mailing distinction and the company-representative names are where identity analysis starts. A residential or virtual-office physical address, a mailing address that does not match, a phone number or representative name that recurs across many otherwise-unrelated USDOT records: these are the raw material of fraud and chameleon-carrier detection. A scraper returns these as plain strings. The signal lives in the relationships between them across the whole population, which a single-record pull cannot see.
The temporal skeleton (MCS150_DATE, ADD_DATE, last-change date, inactivation date)
The record stamps when it was added to MCMIS, when the last MCS-150 was filed, when it was last changed, and, if applicable, when it was inactivated.
The trap: most consumers never read these, and they are the keys to everything else. Every operational field's credibility is a function of how old it is, and these dates are how you measure that. A power-unit count attached to a five-year-old MCS150_DATE is not the same datum as the same count filed last month. Ignoring the date fields is treating a snapshot as if it were live.
Identity and lifecycle flags (POINTNUM, NOTIFY, USDOT_REVOKED_FLAG)
Three quieter fields carry outsized weight. POINTNUM points from a secondary record to the primary USDOT record it has been merged or superseded into, which is how you avoid treating a retired duplicate as a live carrier. The notify flag is set when a company is newly added or reactivated, then reset once a notification letter is generated. The revoked flag marks a revoked new-entrant registration.
The trap: a reactivation flag, a newly added record sharing an address or representative with a recently deactivated one, or a fresh USDOT number that appears shortly after another went out of service near the same address are exactly the patterns that distinguish a reincarnated bad actor from a new legitimate entrant. These fields are invisible to a name-or-DOT lookup and only become signal when you can traverse the full population over time.
The four problems raw data cannot solve
Step back from the fields and the work resolves into four problems. Each is the reason interpreted intelligence sits above raw extraction.
1. Staleness and data decay. The biennial cadence guarantees that a large fraction of records describe a stale reality, and the mileage-year stamp guarantees that even "fresh" filings can carry old operational numbers. Solving this means age-weighting every field against its own date stamp and modeling the decay curve for each field type, because a phone number ages differently than a power-unit count. A raw pull cannot do this; it does not even surface the dates in a usable form.
2. Self-report bias. Power units, mileage, drivers, and cargo are claims, not measurements. The only way to correct for this is cross-validation against independent signals: inspection and crash activity, authority and insurance filings, observed behavior in adjacent datasets. A single record cannot check itself.
3. Identity and chameleon carriers. The hardest and highest-value problem. Reincarnated carriers, shell registrants, and fraud rings reveal themselves only through relationships, shared addresses, phones, representatives, and suspicious lifecycle timing, across the entire population over time. This is graph and entity-resolution work. A scraper that returns one record at a time is structurally incapable of seeing it.
4. Status versus authority versus insurance. Census status, operating authority, and insurance on file are three different questions answered by three different files. Conflating them, which raw single-source data invites, is how brokers tender freight to carriers that look active but cannot legally or financially back the load.
What an interpretation layer actually does
The difference between raw MCS-150 data and carrier intelligence is a set of transforms that a marketplace scraper does not perform:
-
Normalization. Reconciling the letter-bucketed and exact representations of fleet size, standardizing addresses, parsing the equipment and driver sub-fields into usable structure.
-
Temporal modeling. Attaching an age and a confidence to every field based on its date stamp, and decaying each field type on its own curve.
-
Cross-source reconciliation. Joining census data to authority, insurance, inspection, and crash files so each self-reported claim can be checked against something independent.
-
Entity resolution. Linking records across USDOT numbers, addresses, phones, and representatives to collapse duplicates, follow POINTNUM chains, and surface chameleon patterns.
-
Confidence scoring. Returning not just a value but a defensible estimate of how much to trust it, which is the thing a decision actually needs.
This is the layer AlphaLoops is built around, and it is the layer that does not come out of a scraper. The raw field is a commodity. The judgment about the field is not.
What Motus changes in 2026
The mental model behind everything above, the MCS-150 as a periodic form that produces a snapshot, is now shifting under the industry's feet, and anyone interpreting this data needs to track it.
FMCSA is replacing its decades-old registration platform with Motus, a single online dashboard for registration actions. The Unified Registration System is being taken permanently offline, and registration functions in the legacy FMCSA Portal were retired in May 2026, with a short migration blackout around the cutover. The rollout is phased: supporting companies such as BOC-3 filers and insurers got limited access first, with broader access to all regulated entities through 2026. FMCSA continues to accept legacy MCS-150 and OP-1 paper filings during the transition, but paper review runs slow, on the order of a minimum of eight business days, and the long-run intent is for Motus to become the only path.
The consequential change for data interpretation is conceptual. Under Motus, the biennial update becomes a direct edit to a continuously maintained company account rather than the submission of a discrete form, with real-time validation, stronger identity verification, and fraud-prevention checks at the point of entry. The data is meant to move from a series of periodic snapshots toward a continuously validated profile.
Two implications follow. First, the historical data, the billions of field-rows already in MCMIS, was captured under the old snapshot regime and still has to be interpreted with all the staleness and self-report caveats above; Motus does not retroactively clean it. Second, as validation moves to the point of entry, the nature of the errors changes. Outdated, mismatched, or incomplete records that slipped through the old system become easier to spot, which raises the value of an interpretation layer that can compare a carrier's pre-Motus and post-Motus records and flag what changed. The form-era playbook does not transfer cleanly, and treating Motus-era data with snapshot-era assumptions will produce its own new class of mistakes.
A practical reading checklist
When you look at any MCS-150 record, before you act on it, ask:
-
How old is this? Read MCS150_DATE and the last-change date, not just the values.
-
How old is the mileage specifically? Read MCS150MILEAGEYEAR next to MCS150_MILEAGE, always.
-
Is this even a carrier? Read operation and classification; screen out registrants, shippers, and private fleets.
-
What does the fleet number actually mean? Read the OWN / TRM / TRP split, not just TOT_PWR.
-
Is this record the primary one? Check POINTNUM for merges and supersessions.
-
Active status, or active authority? These are different files and different questions. Do not let "Active" stand in for "can legally haul my freight, insured, today."
-
Does anything about the identity recur? Address, phone, and representative reuse across records is where fraud lives.
Every one of these questions is answerable only by reading fields in relationship to each other and to time. That relational, temporal reading is the work. The raw data is just the input.
Primary sources and further reading
-
FMCSA, MCMIS Catalog: Census File Data Element Definitions (the authoritative field-level dictionary): fmcsa.dot.gov/registration/mcmis-catalog-census-file-data-element-definitions
-
FMCSA, Form MCS-150 and Instructions: fmcsa.dot.gov/registration/form-mcs-150-and-instructions-motor-carrier-identification-report
-
FMCSA Registration Modernization Resources Hub (Motus): fmcsa.dot.gov/registration/resources-hub
-
FMCSA Safety Measurement System methodology (for how VMT and power-unit exposure are used)
-
MCMIS census file documentation, FMCSA Volpe Center catalog