You’ve heard it at the fuel desk. You’ve heard it in the comments under every trucking video. Some guy with a flag in his profile picture types the same thing every time: “They’re letting foreigners drive trucks and that’s why the roads aren’t safe anymore.”
It’s a tidy story. Tidy stories are usually the ones worth checking.
Well, somebody finally checked. California’s DMV pulled three years of crash records for nearly 700,000 commercial license holders and lined them up side by side. The result did not go the way the fuel-desk crowd expected.
Let’s walk through what they found, what it actually proves, and – more importantly for your business – what you should be doing about it this week.
What California Actually Did
California DMV’s Research and Development Branch ran what’s best described as a crash record comparison, not a controlled scientific study. The description became public through a multistate court filing and was reproduced in detail by Overdrive.
Here’s the setup:
- 602,257 commercial license holders classified as U.S. citizens
- 69,250 commercial license holders classified as temporary legal residents
- All police-reported crash records from 2022, 2023, and 2024
- 97,960 total crash records, involving 85,394 unique people
Then they divided crash records by the number of license holders in each group and multiplied by 100.
| Crash records per 100 license holders | U.S. citizens n = 602,257 |
Temporary residents n = 69,250 |
Difference |
|---|---|---|---|
| All crashes | 14.809 | 11.915 | 19.5% lower |
| Crashes with at least one injury | 4.908 | 3.947 | 19.6% lower |
| Crashes with at least one fatality or injury | 4.977 | 4.004 | 19.5% lower |
| Crashes with at least one fatality | 0.123 | 0.092 | 25.2% lower |
Every single row went the same direction. The group everybody’s been told to worry about had fewer crash records per licensed driver, including fatal ones.
One wording thing matters a lot here. These are crash records per 100 license holders across three years combined. They are not yearly rates, and they are not “the percentage of drivers who crashed.” Some drivers show up more than once. And a “fatal crash” means somebody died in that crash – not that the CDL holder died.
Now Let Me Take Some Air Out of It
Here’s where I’m going to disappoint anyone hoping to screenshot the table and dunk on their uncle at Thanksgiving.
This comparison does not prove that immigrants are better drivers. It can’t. Here’s why.
It doesn’t count miles. The bottom of the math is license holders, not miles driven or hours behind the wheel. A full-time OTR owner-operator running 120,000 miles a year counts as one. So does someone whose CDL quietly expired two and a half years ago and hasn’t touched a truck since. If one group simply drives less, its crashes-per-license number drops without its crashes-per-mile risk changing at all.
It doesn’t say who caused anything. “Police-reported crash” means the driver was involved. Involvement and fault are two different animals, every owner-operator who’s been rear-ended at a light already knows this. The data also doesn’t confirm the person was even in a commercial vehicle at the time.
It doesn’t adjust for anything. No controls for age, experience, CDL class, route type, urban vs. rural, carrier, equipment, or how hard enforcement runs in a given area. No confidence intervals. No statistical tests.
“Temporary resident” isn’t the same label as “non-domiciled CDL.” California is using a DMV legal-presence category. The federal rule uses a credential and domicile category. They probably overlap a lot. “Probably overlap a lot” is not the same as “are the same thing.”
And it says nothing about Latino drivers specifically. There’s no ethnicity field in that table. BLS data shows 24.3% of “driver/sales workers and truck drivers” identified as Hispanic or Latino in 2024, but that category isn’t limited to CDL holders, and it tells you nothing about domicile.
So what can you take from it? This, and it’s still a big deal:
A large state dataset looked for the pattern everyone assumed was there – and didn’t find it. Before any adjustments, the numbers pointed the other way.
That’s enough to kill the presumption. It’s not enough to flip it.
Here’s the Part Nobody Reads: FMCSA Agrees on the Evidence
This is the detail that should stop the argument cold.
In its September 2025 interim rule, FMCSA said it did not have sufficient evidence from well-designed, rigorous quantitative analyses to reliably show a measurable relationship between a driver’s nation of domicile and U.S. safety outcomes. The agency described its own crash examples as illustrative — not a statistical sample.
So what’s the 2026 final rule actually built on? Two things:
- States often can’t pull a foreign applicant’s complete driving history the way they can pull a domestic one through CDLIS.
- Audits found real state licensing sloppiness — credentials issued past immigration-document expiration dates, wrong credential categories, weak SAVE verification, bad recordkeeping.
Those are legitimate paperwork problems. Fix them. But notice what they are not: an expiration-date mismatch in a DMV file does not mean a driver failed a road test, blew a pre-trip, or caused a wreck. Somebody in an office typed the wrong date. That’s a filing error, not a safety record.
Confusing those two things is how this whole debate went sideways.
The Category Pile Up
Most of the noise online comes from mashing together six different things that have nothing to do with each other. If you only remember one section of this post, make it this one:
| Term | What it actually means | What it does not tell you |
|---|---|---|
| Latino / Hispanic | Ethnicity | Citizenship, status, CDL type, English, or safety |
| Citizenship | Legal nationality | Ethnicity, domicile, or driving skill |
| Work authorization | A federal legal classification | Whether you qualify for a CDL under every rule |
| Domicile | Your legal home / main residence | Race, ethnicity, or lawful presence |
| Non-domiciled CDL | A credential category tied mainly to foreign domicile | Undocumented status, Latino identity, bad English, or unsafe driving |
| English proficiency | Ability to do the safety-critical tasks in 49 CFR 391.11(b)(2) | Citizenship, ethnicity, accent, or whether you speak Spanish |
That English rule applies to every interstate driver, period. A citizen can fail it. An immigrant can pass it. The test is whether you can read signs and communicate with officials, not whether you have an accent. We broke down how that’s playing out at roadside in our post on English requirements and CVSA enforcement, and how state testing changed in Texas’s English-only CDL exam.
If you’re Latino, a citizen or green card holder, and domiciled in your state, this rule is not about you. Your name, your accent, and your Spanish playlist don’t make you a non-domiciled driver. Nobody, not a broker, not an inspector, not a dispatcher, gets to guess your credential from your last name.
What It Costs the Guy Who Actually Is Affected
Now the harder half.
The final rule took effect March 16, 2026. If you followed the original 2025 crackdown, the shape is familiar, the remaining foreign-domiciled categories are narrow: H-2A, H-2B, and E-2. You need an unexpired passport and I-94, SAVE verification, an in person visit, and the credential runs one year max – or until your immigration document expires, whichever hits first. Employment Authorization Documents are no longer accepted for this.
Read that again, because it’s the part that catches people: lawfully present does not automatically equal CDL-eligible. DACA recipients, TPS holders, refugees, asylees – people who are legally in the country and legally allowed to work – can still fall outside those three categories.
FMCSA’s own estimates: roughly 200,000 non-domiciled CDL holders, about 6,000 new issuances a year under the new criteria, and roughly 194,000 people exiting commercial driving over as long as five years.
Picture what that means for one truck. I’ll give you a composite – this isn’t one specific person, it’s the same story I’ve heard versions of a dozen times. Rafa, twelve years driving, zero preventables, DOT number in good standing, a 2019 Cascadia with about $1,900 a month left on it and a reefer trailer he just paid off. He goes to renew. He walks out without a license.
The truck payment doesn’t care. Neither does the insurance, the parking, the permits, or his daughter’s tuition. His MC authority stays active — but authority doesn’t drive the truck. FMCSA’s cost analysis illustrates a roughly $16,000/year wage gap between CDL and non-CDL work, then largely sets it aside on the assumption affected drivers find similar work. Nothing in that analysis models truck equity, a forced sale, or a loan default. It is not an owner-operator balance sheet. If you want to see how brutal those fixed costs are when revenue stops, run your own numbers through our cost-per-mile calculator ,the fixed-cost column doesn’t blink.
And there’s a second door closing. Some carrier-screening platforms have started converting credential attributes into freight access. Highway rolled out an optional broker screen tied to the non-domiciled status of a carrier’s primary account owner. The company framed it as a theft-pattern signal, not a crash-risk finding. Fair enough – but a carrier can get declined without a single safety violation anywhere in its file. That’s a business problem wearing a safety costume, and it’s the same fight we covered in getting a bad FreightGuard report removed.
Where the Fight Stands Right Now
The rule is being challenged in Rivera Lujan v. FMCSA (D.C. Circuit No. 26-1032, consolidated with King County’s No. 26-1046). On May 5, 2026, a divided panel denied an emergency stay and put the case on a fast track. That kept the rule running, it was not a ruling on the merits. Oral argument is set for September 15, 2026.
Politically it’s near-even: 21 states plus D.C. oppose the rule, 23 support it.
Two claims flying around that you should handle carefully:
- “Fewer than 50 of California’s ~65,000 prior non-domiciled holders can renew.” That’s an estimate in the states’ amicus filing, not an audited federal count.
- “Unauthorized immigrants hold up to a third of all CDLs.” No identified national dataset behind that one, and it swaps “unauthorized immigrant” for the completely different federal category “non-domiciled CDL holder.” FMCSA’s own figure is around 5%.
Both sides are stretching. Know which numbers you can defend.
The Final Mile
The California data isn’t a trophy. Don’t treat it like one.
What it is, is a big administrative dataset that went looking for a pattern everybody assumed was there, and came up empty. Before any adjustment, temporary-resident CDL holders showed up in fewer crash records per license than citizens, across every severity level measured. FMCSA’s own record admits it lacks rigorous evidence tying domicile to crash outcomes.
That means this argument should be fought where it belongs: on licensing compliance, verifiable driving history, individual qualifications, fair process, and accurate records.
Latino identity. Speaking Spanish. Being born somewhere else. Having an accent. Holding a non-domiciled CDL. Five different things. Not one of them is a substitute for looking at how a specific driver actually runs.
Your MVR is a better argument than anybody’s opinion. Keep it clean, keep it handy, and make them look at it.
Frequently Asked Questions (The Stuff You’re Probably Still Wondering)
1. Do non-domiciled CDL drivers cause more accidents than U.S. citizens?
2. What did the California DMV crash data actually show?
3. Does this prove immigrants are safer drivers?
4. What is a non-domiciled CDL?
5. Who can still get a non domiciled CDL in 2026?
6. Can DACA, TPS, refugee, or asylum holders still get a CDL?
7. Does being Latino or Hispanic affect my CDL status?
8. Is the non domiciled CDL rule still in effect?
9. What should I do if my CDL is being downgraded?
10. Can a broker refuse my loads over non-domiciled status?
