FMCSA Random Drug Testing Rate 2026: 50%, and How the Pool Math Actually Works

Published September 18, 2026 | TruckerNavi Inc | 11 min read

By the TruckerNavi compliance team, led by Dmitry Borovoy, a New Jersey-licensed insurance producer and author of "Safety Manager: From Zero to Expert".

The short answer: for calendar year 2026 the FMCSA minimum random testing rates are 50 percent for controlled substances and 10 percent for alcohol. The rates did not change from the previous year, and because they did not change, FMCSA was not required to publish a notice in the Federal Register. The current table for every DOT agency lives on the Department of Transportation's Random Testing Rates page, which lists FMCSA at 50 percent and 10 percent for 2026 and notes that the rate last changed in 2020.

Both numbers come from 49 CFR 382.305(b), and both are percentages of the average number of driver positions. That phrase is where most owners go wrong: it is not the number of drivers on your payroll today, and it is not the number on December 31. It is an average across your selection periods, defined precisely in 382.305(j)(1).

Everything below is sourced to the rule text, to DOT's Office of Drug and Alcohol Policy and Compliance, and to FMCSA's own published answers. This is a walk-through of the rules, not legal or medical advice, and it does not predict how any particular investigator will treat any particular file.

What is the FMCSA random drug and alcohol testing rate for 2026?

Fifty percent for drugs. Ten percent for alcohol. Both are floors, not targets, and both apply for the whole calendar year 2026.

The mechanism that sets them is worth understanding, because it is why you should check the rate every January rather than assume it:

• The Administrator moves the rate, not Congress. Under 382.305(f), the FMCSA Administrator's decision to increase or decrease the controlled substances rate is based on the reported positive rate for the entire industry, drawn from the Management Information System reports required by 49 CFR 382.403. The alcohol rate moves the same way under 382.305(c), based on the industry violation rate.
• A change gets published, and takes effect the following January 1. The rule says that in the event of a change the Administrator will publish the new minimum annual percentage rate in the Federal Register, and that the new rate applies starting January 1 of the calendar year following publication.
• No change means no notice. This is the detail that confuses people every winter. If the rate stays put, there is nothing to publish. DOT's rates page states plainly for 2026 that the random rates did not change and that FMCSA is therefore not required to publish a Federal Register notice. Silence in December is not a hint that the rate dropped.

The ladder itself is written into the rule, so you can see where the next move would come from:

Note the asymmetry: lowering the rate is discretionary and needs two clean years, raising it is mandatory and needs one bad year. Plan your budget around 50 percent staying put.

How many random tests does my fleet actually owe this year?

Multiply your average number of driver positions by 0.50 for drug tests and by 0.10 for alcohol tests. DOT states the arithmetic in the plainest possible terms in its guidance: an employer with 100 safety-sensitive employees would have to ensure that 50 or more random drug tests and 10 or more random alcohol tests were conducted during the calendar year.

Because the rate is a minimum and tests come in whole numbers, the answer is to round up. The text of 382.305 states no rounding rule, but FMCSA does it in its own published example, turning 3.5 required tests into 4. And 2.5 tests performed as 2 puts you below 50 percent, which is the only outcome the rule forbids. Over-testing is always permitted.

Two things fall out of that table that surprise small carriers. First, the alcohol obligation is tiny and still mandatory. At six drivers you owe less than one alcohol test on paper, which is exactly why so many small fleets do zero and cannot explain it. Do one. Second, a three-truck operation is not exempt from anything. The percentage scales down; the obligation does not disappear.

What is the "average number of driver positions", and why is it not my December headcount?

Because the rule defines the calculation, and the definition is an average. 49 CFR 382.305(j)(1) says it directly: to calculate the total number of covered drivers eligible for random testing throughout the year, you must add the total number of covered drivers eligible for testing during each random testing period for the year and divide that total by the number of random testing periods.

Two more sentences in the same paragraph carry real weight:

• "Covered employees, and only covered employees, are to be in an employer's random testing pool, and all covered drivers must be in the random pool." Both halves are enforceable. Leaving a driver out is a violation, and putting a non-covered person in is a violation.
• If you run random testing more often than once a month, you do not have to recompute the number more than once a month.

Worked example, four quarterly selection periods:

Now look at the three wrong answers the same carrier could have given:

• Used the January number (4). Owner schedules 2 drug tests. Actual requirement is 3. He is one test short and will not find out until an investigator recomputes it.
• Used the December 31 number (6). By coincidence this gives the right answer here. Change the story so the fleet ends the year at 3 drivers after a bad Q4, and the same method produces 2 tests against a requirement of 3.
• Used the peak (8). Four tests instead of three. Legal, since over-testing is allowed, but he paid for a test he did not owe and never knew it.

The average is the only method the rule recognises. Write the eligible headcount down on the first day of every selection period, and the calculation does itself in December.

FMCSA has published the formula outright, in the answer covering carriers that operate for only part of a year or whose driver count swings hard. Use T = 50% × D/P for controlled substances and T = 10% × D/P for alcohol, where T is the number of tests required, D is the total number of drivers subject to testing across the periods, and P is the number of selection periods in a full calendar year. For any period in which you were not yet subject to 382.305, enter a zero.

FMCSA's own worked example is the one to copy. A carrier starts operating in August and tests quarterly, so P = 4. It has 16 drivers subject to testing in the third quarter and 12 in the fourth. D = 0 + 0 + 16 + 12 = 28, D/P = 7, and T = 50% of 7 = 3.5, which must be rounded up to 4. The carrier owes 4 controlled substances tests between its first day of operation in August and the end of the year, and following the requirement to spread testing reasonably through the year, two should fall in the third quarter and two in the fourth. Note that FMCSA rounds up and enters zeros for the periods before you existed. A carrier that starts in August does not get to divide by 2.

Who must be in the random pool, and who must never be in it?

The pool is not "everyone who works here" and it is not "the drivers I trust least". It is a defined population.

Must be in:

• Every CDL driver who operates a commercial motor vehicle subject to Part 382, at all times.
• Part-time drivers. FMCSA answers this one directly: all CDL drivers of CMVs must be included in the DOT random pool at all times. A driver who covers two runs a month is in the pool for the whole month.
• The owner, if the owner drives. DOT is explicit that people are chosen based on their safety-sensitive job function, not their occupational title. Supervisor, volunteer, contractor, owner-operator: if the function is covered, the person is in.
• Drivers who work for two carriers. FMCSA states that a driver must be in the DOT random pool of each employer he is employed with. Both pools, both sets of odds.

Must not be in:

• Non-CDL drivers. FMCSA's answer is unambiguous: an employer may not include non-CDL drivers in the DOT random testing pool, even those operating heavy vehicles, and non-DOT covered employee testing must be placed in a pool completely separate from the DOT covered pool. The underlying rule is 49 CFR 40.347(b)(2): employees not covered by DOT agency regulations may not be part of the same random pool with DOT covered employees. What a compliant program contains end to end is laid out in the FMCSA drug and alcohol program overview.
• Office staff, mechanics who do not drive CMVs, dispatchers. Same reason. You may test them under your own company authority, but not as a DOT test and not out of the DOT pool.
• Drivers who have left. DOT's stated best practice is to refresh the pool just before performing a random selection, so that it contains every covered employee subject to DOT random testing and excludes those who are not.

Do I need a consortium, or can I run my own pool?

It depends on one number: how many CDL drivers you have.

• More than one CDL driver: FMCSA states that the employer may manage its own random testing program. May, not must. Most small carriers still outsource it, because the selection records and the calendar are the hard part, not the arithmetic.
• Exactly one CDL driver and not leased to another motor carrier: FMCSA states such companies are required to place the driver in a consortium for random testing purposes. A one-person pool cannot produce a random selection, so the rule sends you into a larger one.

Using a service agent does not transfer the liability. 382.305(j)(2) says you may use a consortium or third-party administrator and your drivers may be part of a larger pool, however you must ensure that the service agent is testing at the appropriate percentage established for your industry and that only covered employees are in the pool. When the investigator finds a pool that tested at 38 percent, the finding is against your DOT number, not against the TPA's. Picking one and verifying it properly is its own subject, covered in how to choose and verify a consortium or TPA. If you would rather not shop, our Drug & Alcohol program bundles consortium enrolment, random selections and Clearinghouse handling at $150 a year.

One more combination rule worth knowing if you run mixed operations. Under 382.305(o), an employer subject to more than one DOT agency's rules may either keep separate pools by rate, or select everybody at the highest minimum rate any of those agencies sets. And under 382.305(n), an individual employee subject to two agencies is tested at the rate of the agency regulating more than 50 percent of that employee's function.

How are drivers actually selected for a random test?

49 CFR 382.305(i) sets three requirements, and every one of them is checkable from your records:

1. (i)(1) A scientifically valid method. The rule names the acceptable ones: a random number table, or a computer-based random number generator matched with drivers' Social Security numbers, payroll identification numbers, or other comparable identifying numbers.
2. (i)(2) An equal chance every time. Each driver selected under the process must have an equal chance of being tested each time selections are made. Not once per year. Each time.
3. (i)(3) Tested inside the period. Each driver selected for testing shall be tested during the selection period. This is the sleeper violation, and it gets its own section below.

DOT's guidance also names methods that are not acceptable, and the list is worth reading out loud to anyone who thinks this is a formality: selecting numbers from a hat, rolling dice, throwing darts, picking cards, or selecting ping pong balls. A spreadsheet where the owner types in a name is in the same category.

If you have several terminals, you may select by location, but not casually. FMCSA states that a location-based selection requires a two-stage process: first select the location or locations, then select the employees at the selected location. The selections must still ensure that each employee in the pool has an equal chance of being selected and tested no matter where the employee is located.

Why did the same driver get pulled twice while another was never picked?

This is the single most common complaint in a small fleet, and it is usually evidence that the program is working correctly.

DOT addresses it head on: in a truly random selection process, a high probability exists that some employees will be selected several times while others may never be selected, because after each selection the employee's name is returned to the same pool and he becomes just as likely as anyone else to be selected next time.

Put numbers on it. Take a six-driver pool with four selections in a year, one driver drawn each time, each draw independent with the name returned afterwards:

Those are the plain odds of the model, not a regulatory figure, but they are the answer to the argument in the parking lot. Roughly one driver in eight in a small pool gets pulled twice in a year purely by chance, and roughly half get pulled not at all.

What does "spread reasonably throughout the year" actually require?

382.305(k)(2) requires the employer to ensure that the dates for administering random tests are spread reasonably throughout the calendar year, and 382.305(k)(1) requires them to be unannounced. Those two work together: a predictable schedule is not really unannounced.

DOT's guidance puts a cadence on it. Selections and testing should be performed at least quarterly, and some employers select more frequently, which DOT calls a good idea. The stated best practices are to spread testing dates reasonably throughout the year in a non-predictable pattern, and to vary the time of day: start, middle or end of a shift. DOT even writes down the failure it is trying to prevent, the moment drivers say that the last Friday of every month the second shift gets tested.

Here is what a compliant year looks like for a fleet averaging six covered drivers, owing 3 drug tests and 1 alcohol test:

Note what that schedule buys you. One test per quarter on a six-driver fleet produces four tests where three are required, which is 67 percent against a 50 percent floor. That extra test is not waste. It is your margin for a cancelled collection, a driver who quits between selection and collection, or a lab problem in December.

The opposite pattern, three collections crammed into the last two weeks of December, hits 50 percent exactly and fails 382.305(k)(2) on its face. The percentage and the distribution are two separate requirements, and an audit checks both.

What must a driver do the moment he is notified?

382.305(l) requires the employer to ensure that a driver notified of selection proceeds to the test site immediately. If the driver is performing a safety-sensitive function other than driving a commercial motor vehicle at the time of notification, the employer must instead ensure that the driver ceases to perform that function and proceeds to the testing site as soon as possible.

DOT's guidance leaves no wiggle room on the word: immediately does not mean two hours, and after notification all of the employee's actions must lead to an immediate specimen collection. Two more points from the same guidance:

• There is no second attempt. Once the employee is notified to report for testing and the test does not occur, the opportunity for that random test is over. There is no second bite of the apple. You do not get to notify him again next week and call it the same selection.
• Notification must be discreet. Random testing has to be conducted in strict confidence with a limited number of people knowing the selection list, precisely to preserve the element of surprise.

Alcohol testing carries its own timing limit that has nothing to do with convenience. 382.305(m) allows a driver to be tested for alcohol only while performing safety-sensitive functions, just before performing them, or just after ceasing to perform them. FMCSA has answered what happens when an employer ignores this: a driver sent for a random alcohol test while off duty and not performing a safety-sensitive function was improperly tested, the employer must cancel the test and document the circumstances, and the driver does not enter the return-to-duty process even on a positive result. The thresholds and consequences on a properly conducted alcohol test are laid out in alcohol testing thresholds and what each number triggers.

What happens if the driver simply does not show is not a scheduling problem, it is a refusal, and a refusal carries the same consequences as a positive. That path, including the failure of an owner-operator to appear when called by a consortium, is worked through in why a refusal counts as a positive.

What if the selected driver is on vacation, sick or laid off?

You do not simply skip him, and you do not automatically swap him either.

FMCSA's answer is narrow: an alternate driver may be selected only if the primary driver selected will not be available for testing for the entire selection period because of a long-term absence due to layoff, illness, injury, vacation or other circumstances. A driver who is out on a three-day run is not unavailable for the quarter. He is unavailable today.

DOT's best practices fill in the rest:

• If the selected employee is known to be unavailable for the whole cycle, document the reason and make up the shortfall by making another selection, or an extra selection in the next cycle.
• If the employee is selected on his day off and has not been notified, test him on his next shift within the same selection cycle.
• No employee should be excused because of operational difficulties. "The load was hot" is not a reason.

And the hard edge, from 382.305(i)(3): the driver shall be tested during the selection period. FMCSA has confirmed what happens when he is not. Asked about a driver selected in the first quarter but tested on April 4, FMCSA answered that the motor carrier is in violation because the driver was not tested in the selection period in which he was selected, while noting that the test still counts toward the carrier's annual minimum. You keep the credit and you keep the violation.

Which tests count toward the rate, and which do not?

A surprising number of carriers reach December believing they are at 50 percent, then discover they are at 33.

The cancelled-test line is the one that quietly sinks small fleets. If you owe three drug tests, run exactly three, and one comes back cancelled, you finished the year at two out of three. On the alcohol side 40.273(c) requires the technician who determines a cancellation to inform the designated employer representative within 48 hours, so you will know in time to replace it, provided somebody is watching the mailbox. On the drug side the timing is looser but the arithmetic is identical, which is why the only safe habit is to check every selection against a completed, uncancelled result rather than assume.

That is the practical argument for scheduling above the minimum. DOT's own guidance says that if you think you might not meet your annual testing rate requirement, increase your testing.

How carriers actually fail the random testing check on a DOT audit

An investigator reviewing this area is not looking for a story. He is looking for a number, a calendar and a stack of selection records. There are six standard ways to lose:

1. Shortfall. Tests conducted came out below 50 percent or below 10 percent of the average number of driver positions. Usually caused by computing the average wrong, not by skipping tests deliberately.
2. Bunching. The percentage is met, but every collection sits in one or two months. Straight 382.305(k)(2), and it is visible in ten seconds from the collection dates.
3. No selection records. The carrier has the negative results and nothing showing who was drawn, when, from what pool, by what method. Without the selection list, there is no way to prove the tests were random rather than chosen.
4. Wrong pool contents. A covered CDL driver was never in the pool, or a non-covered person was. Both contradict 382.305(j)(1).
5. Selected but not tested in the period. Names appear on a selection list with no matching collection inside that quarter, and no documented reason.
6. Delay after notification. The notification log shows 9:15 a.m. and the collection shows 2:40 p.m. with a load delivered in between.

It is worth knowing how heavily each of those lands. FMCSA classifies violations for safety fitness purposes in Appendix B to Part 385, and this area appears twice:

The distinction matters. Having no random program at all is the acute one. Having a program that came in under the percentage is critical. Neither is a paperwork nicety, and both are computed from records you either produced or did not.

DOT's counter-measure is boring and effective: document everything on the entire random testing process. That means the numbers, the names drawn, the dates and times of notification, the dates and times of collections, and the reason a selected employee was not tested during a cycle. DOT's phrasing is "if you're not sure, document it".

The other stated best practice is not to reconcile at the end. DOT tells service agents and C/TPAs to monitor completed tests against selections on an ongoing basis, daily or weekly, and to chase any test not completed within a day or a week of the expected date. Employers and service agents, it says, should not wait until the end of the selection period to reconcile the random testing numbers, calling that a weak business practice.

What records must you keep, and for how long?

49 CFR 382.401 sets the retention schedule, and requires all of it to be maintained in a secure location with controlled access.

Separately from the selection file, every random result that comes back positive or is recorded as a refusal has to be reported into the FMCSA Clearinghouse, and each of your drivers needs an annual query on file. The registration and query mechanics are in the FMCSA Clearinghouse registration guide.

Read the last line again. Two business days. If your selection records live only inside a TPA portal you have never logged into, that clock is going to hurt. Download the selection list after every draw and keep a copy where you can find it.

One related obligation people forget entirely: 382.403. You must prepare and maintain an annual calendar year summary of your testing results when requested by DOT, a DOT agency, or a state or local official with regulatory authority. If FMCSA notifies you during the month of January, you must prepare and submit the MIS report so that it is received by March 15 of that year. A service agent may prepare the report, but a company official must certify its accuracy and completeness, no matter who prepares it.

A 12-month routine that keeps the percentage honest

1. Every January, check the rate. Look at the DOT rates page rather than assuming. No Federal Register notice means no change, not permission to guess.
2. Write down the eligible headcount on day one of every selection period. Four numbers a year is the whole calculation under 382.305(j)(1).
3. Refresh the pool immediately before each draw. Add new hires, remove departures, and confirm nobody non-covered has crept in.
4. Select at least quarterly, and vary the day and the shift. Monthly is better. Last Friday of the month is the pattern DOT names as a failure.
5. Schedule above the minimum. One extra drug test a year absorbs one cancelled collection without pushing you under 50 percent.
6. Notify discreetly and send the driver immediately. Log the time of notification and the time of collection. Those two timestamps are the whole defence on 382.305(l).
7. Reconcile every month, not in December. Selections drawn against collections completed, with a written reason for every gap.
8. Download and keep the selection lists yourself. Two business days is not enough time to start a support ticket.
9. Keep the alcohol tests on the calendar. Ten percent of a small fleet rounds to one test, and one test is not zero.
10. Never let a supervisor pick a name. Hat, dice, darts, cards and ping pong balls are named as unacceptable; a typed name is worse than all of them.

None of this is complicated. It is a calendar, a headcount and a folder. What makes it expensive is that all three failures are invisible until an investigator recomputes your average and lays your collection dates on a timeline.

FAQ

Related articles

• Refusal to Test: Why It Counts the Same as a Positive
• Post-Accident Testing: When It Is Required and the Clock You Are On
• Alcohol Testing Thresholds: 0.02, 0.04 and What Each One Triggers
• How to Choose and Verify a Consortium or TPA
• Supervisor Drug and Alcohol Training: The 60+60 Rule
• FMCSA Clearinghouse Registration: Queries, Reporting, Deadlines
• FMCSA Drug and Alcohol Program: What a Compliant Program Contains
• Drug & Alcohol Program, $150 a year: consortium, random pool, Clearinghouse
• CSA Scores Explained: BASICs, Thresholds, Methodology