What Is a CSA Score — and Which Hire Moved Yours?

A CSA score is not one number. It is a set of percentiles — one per category — that rank your violation rate against carriers with similar exposure, built from roadside inspections and crashes recorded under your DOT number over the past two years. CSA stands for Compliance, Safety, Accountability, the FMCSA program; the scores themselves come out of its Safety Measurement System (SMS). A percentile runs 0 to 100, and higher is worse: 100 marks the highest violation rate in your peer group (GAO-14-114).

Two things follow from that design, and both surprise carriers.

It is a ranking, not a grade. You are not measured against a fixed standard. You are measured against carriers with a similar number of inspections, which means your percentile can climb in a month when you did nothing worse than usual — because the peer group got cleaner.

It is not your safety rating. The Satisfactory / Conditional / Unsatisfactory rating is a separate instrument with its own legal consequences, and confusing the two is how carriers panic about the wrong document. That distinction is worked through further down.

How do I check my CSA score?

Carriers view their own scores by logging in to FMCSA’s Safety Measurement System at ai.fmcsa.dot.gov/SMS with a USDOT number and PIN. Complete results are available to the carrier itself and to enforcement users; the public view is narrower.

Look past the percentile on the front page. The number tells you something happened. Only the underlying inspection and violation records tell you what, when, and to whom — and that detail is what the rest of this article works from.

Which hire moved it?

Most carriers arrive here the same way: the score turned the wrong way this month, nothing changed in the shop or in dispatch, and the only real variable last quarter was a handful of new drivers.

That guess is usually right, for a structural reason. GAO reported in February 2014 that among the carriers it assessed, 93 percent owned or operated 20 or fewer motor vehicles, and that under FMCSA’s data sufficiency standards a one-truck carrier with two inspections carrying unsafe driving violations is not even scored — but “a single additional violation, regardless of the severity of the violation, would likely mean that the carrier would be scored above threshold and prioritized for intervention” (GAO-14-114). At that size your score is not a fleet average. It is a short list of events, and each one has a driver’s name on it.

What follows is not a line-by-line walkthrough of the SMS formula. It is enough of the mechanism to know which lever you are holding, then a method for working backward from a moving score to the hire behind it.

Which parts of your CSA score are actually a hiring problem?

Most of the categories track things a person does behind the wheel; the rest mostly track what leaves your yard. The Safety Measurement System sorts roadside inspection violations into six Behavior Analysis and Safety Improvement Categories, plus a seventh category — the Crash Indicator — which measures crash involvement history rather than behavior at roadside (GAO-14-114). Before you spend a quarter chasing the wrong department, split them. The table describes the categories as currently displayed; FMCSA has approved a reorganization that renames and regroups several, covered further down.

Two of those rows get misfiled constantly.

Driver Fitness is an office failure disguised as a driver failure. An expired medical certificate found at roadside is not risky driving. It is a qualification file nobody audited. The obligation is already in the rulebook: under 49 CFR 391.25 you must request the driving record of each driver at least once every 12 months, covering at least the preceding 12 months, from every state where the driver held a CMV operator’s license or permit, review it, and note in the qualification file who did the review and when (49 CFR 391.25). A carrier that runs that review on schedule rarely has a Driver Fitness problem.

Hours-of-Service is shared. Log violations come from drivers, but also from loads that cannot legally be run in the time quoted. If HOS is climbing across several drivers at once, that is dispatch, not recruiting.

Why does one new hire move the number this much?

Because SMS does not score you against an absolute standard — it ranks your violation rate against carriers with similar exposure, and small denominators are violent. GAO’s February 2014 review, describing CSMS Methodology Version 3.0.1 (August 2013), laid out the calculation in three steps: FMCSA “assigns differing weights to each violation that occurred over the past 2 years, depending on the relative severity of each violation and the amount of time elapsed between the violation’s occurrence and the score’s calculation,” sums those weighted violations per BASIC, and divides by a measure of exposure — the time-weighted number of relevant inspections for most BASICs, and vehicles adjusted by miles traveled for Unsafe Driving and the Crash Indicator. FMCSA then sorts carriers into “safety event groups” and, within each group, ranks violation rates into a percentile “ranging from 0 to 100, where 100 indicates the highest violation rate and the highest estimated risk for future crashes” (GAO-14-114).

Three consequences fall out of that design, and all three land on the hiring desk.

One: severity is not uniform. Under the methodology GAO reviewed, reckless driving in the Unsafe Driving BASIC carried a severity weight of 10 out of 10, while unlawful parking in the same BASIC carried 1 out of 10. That spread is being flattened: the approved enhancements move severity weighting from the 1-to-10 scale toward a simplified one. Either way, when you review a candidate’s roadside history, read what the violations were and not only how many there were.

Two: recency compounds. The time weight means a violation from last month counts for more than one from twenty months ago. That cuts both ways: the driver you hired in spring can still be raising your percentile in the fall, and a clean stretch takes time to register.

Three: the cliff is one-way. A carrier can sit below the data sufficiency threshold with no displayed score, then flip into the intervention queue on one violation. GAO added that “a carrier such as this will have a very difficult time improving its SMS score to be below threshold.”

Intervention thresholds differ by category. We are not printing values here — pull the current numbers from FMCSA’s SMS Methodology document rather than trusting a remembered one.

What FMCSA has approved, and what it has not scheduled

FMCSA published “Enhanced Carrier Safety Measurement System (SMS)” in the Federal Register on November 20, 2024, under Docket FMCSA-2022-0066 (full text). Four approved changes reach into everything above:

• The BASICs become compliance categories — “simpler and more relatable terminology.”
• Controlled Substances/Alcohol stops being its own category. Those violations move into a new Unsafe Driving compliance category, from inside which FMCSA will “continue to hold carriers and drivers accountable for drug and alcohol compliance.”
• Vehicle Maintenance splits in two — Vehicle Maintenance, and Vehicle Maintenance: Driver Observed. That second one is the pre-trip column this article keeps pointing at.
• The lookback tightens. FMCSA moves to “calculating percentiles only for carriers with cited violations in the past 12 months” for Hours of Service, Vehicle Maintenance, Vehicle Maintenance: Driver Observed, Hazardous Materials, and Driver Fitness.

What the notice does not give is a start date: “a follow-up notice in the Federal Register will announce the launch date.” Until that appears, the categories in the table above are what you are looking at. Do not put a date in your compliance calendar that FMCSA has not published.

Is a bad CSA score the same thing as a bad safety rating?

No, and confusing the two is how carriers panic about the wrong document. A BASIC percentile is a prioritization number that moves with your monthly data. A safety rating is a formal finding issued after a compliance review, with teeth on a fixed schedule.

Under 49 CFR 385.9, “a safety rating will be issued to a motor carrier within 30 days following the completion of a compliance review.” The rating grid in appendix B to part 385 runs on six factors — General (parts 387 and 390), Driver (parts 382, 383 and 391), Operational (parts 392 and 395), Vehicle (parts 393 and 396), Hazardous Materials, and the Accident Factor — and the arithmetic is blunt: two or more unsatisfactory factors produce an overall Unsatisfactory rating, as does one unsatisfactory factor together with more than two conditional factors (49 CFR part 385, appendix B).

Note what the Driver factor covers. Parts 382, 383 and 391 are the drug and alcohol program, CDL standards, and driver qualification — almost entirely about who you hired and what you documented. The hiring desk owns that factor outright, and its work also shows up in the Operational factor, which runs on parts 392 and 395 and is largely driver conduct too.

The Accident Factor is narrower than people expect. It applies “only when a motor carrier incurs two or more recordable accidents occurred within the 12 months prior to the CR” — below that gate, five factors are live and Driver is one of five. Past it, an urban carrier (operating entirely within a 100 air-mile radius) with a recordable accident rate greater than 1.7 draws an unsatisfactory accident factor; every other carrier, greater than 1.5. Recordable tracks the definition of accident at 49 CFR 390.5T — section 390.5 itself is suspended — a fatality, bodily injury requiring immediate medical treatment away from the scene, or disabling damage requiring a tow, excluding occurrences involving only boarding and alighting, or only loading and unloading cargo (49 CFR 390.5T).

The consequences run on a clock worth knowing by heart. A carrier rated Unsatisfactory that hauls hazardous materials or passengers is prohibited from operating a CMV in commerce “beginning on the 46th day after the date of the FMCSA notice”; all other carriers face prohibition beginning on the 61st day (49 CFR 385.13). A motor carrier “must make a request for an administrative review within 90 days of the date of the proposed safety rating,” and where the proposal is Unsatisfactory the rule says it “should submit its request within 15 days from the date of the notice” so FMCSA can decide before the prohibition attaches (49 CFR 385.15). A corrective-action upgrade request under 385.17 may be filed at any time.

A high BASIC percentile does none of that on its own. It makes a compliance review more likely, which is why you treat it as an early warning, not a scoreboard.

How do you trace a moving BASIC back to a hiring decision?

What the required checks still cannot tell you

They cannot tell you how a driver behaves in the gap between rules. Work Steps 1 through 5 honestly and you will have the MVR, the PSP report, previous employer responses under 391.23, and a Clearinghouse query. Every one records the same kind of thing: an event serious enough that somebody was legally required to write it down — a conviction, a positive test, a roadside violation, a DOT-recordable crash.

The behavior that fills your Unsafe Driving BASIC mostly lives below that line. A driver who runs eight over for two years without getting caught has no MVR entry and no PSP line until the inspection that finally lands in your data. A driver who skips the walkaround has nothing on file until the lamp is out at a scale. A driver who has burned four carriers on abandonment leaves a pattern only those four carriers know about — the same blindness behind the real cost of a bad driver hire.

That gap is what a first-party behavioral layer is built for. On cdlscan, carriers write reviews of drivers they actually employed, so observations that never became reportable events — reliability, communication, how equipment came back, whether the driver finished the contract — still leave a trace. Any carrier can contribute, not only subscribers to a screening database. Our overview of how carrier-written driver reviews work covers the format.

Be clear about what that does and does not do. It will not lower your CSA score; nothing lowers a percentile except time, clean inspections, and correction of records that were wrong in the first place. It improves the decision at the front door — the only lever you control before the violations exist. Run the required checks exactly as written, then look the candidate up as an additional input, or leave a review of your own so the next carrier is not guessing either.

One boundary matters more than the rest. The formal background check is legally required, and this layer sits next to it, never in place of it. If you intend to rely on third-party information about a candidate as a factor in a hiring decision, ask the provider which category of information they consider it under the Fair Credit Reporting Act, and confirm with your own counsel which FCRA obligations attach. That call belongs to a lawyer; the FTC’s employer guidance on using consumer reports is where to start the conversation.

How do you know the change is working?

Watch the raw inputs, not the percentile — it is lagging and relative, and it will mislead you for at least two quarters. Three checks, fastest first.

• Violations per relevant inspection, by hire cohort. Compare drivers hired before your policy change to those hired after. It moves within a quarter and it is yours alone, unaffected by what your peer group did.
• Clean inspection share. Rising is the signal you want, and it is the only thing that eventually pulls a percentile down as old violations age out.
• Concentration. Violations still bunched on one or two drivers means the recruiting screen is the problem. Spread evenly across twenty, it is training and dispatch pressure instead.

If those move and your percentile does not, you are probably not wrong yet. Rank-based scores lag the behavior that feeds them, and thin exposure lags hardest.

Frequently asked questions