How Long Drugs Stay in Your System, and Why That Is Not Impairment

This is one of the most searched questions there is, and almost every page that answers it gives you a number that is wrong. Not slightly wrong. Wrong in a way that matters if you are facing a charge.

There is no number. How long a drug is detectable depends on the dose, how often you use, which body fluid is tested, which molecule the laboratory looks for, and what cutoff that laboratory set. Change the cutoff alone and the answer changes by weeks.

This page cannot tell you when you will test negative, and it is not built to. If you are subject to court-ordered or probation testing, the honest answer is that nobody can give you a safe date, and treating any published range as one is a bad bet with your liberty. What this page can do is show you why a positive test is not proof that you were impaired.

The one thing worth understanding

Impairment is measured in hours. Detection is measured in days, and sometimes in weeks. Those are two different clocks, and a drug test reads the slow one.

That is not a defense lawyer’s framing. It is the position of the forensic toxicology community that advises prosecutors. The National Safety Council’s Alcohol, Drugs and Impairment Division concluded in its 2021 recommendations that urine is an inferior matrix to blood and oral fluid because it may represent historical use or exposure unrelated to observed impairment, and said future versions of its recommendations will not include urine as a recommended matrix at all.

There is a second layer. For several common drugs the molecule the laboratory actually looks for is not the one that affects you. A federal urine test for cannabis targets THC-COOH, which is inactive. It cannot show impairment even in principle. The same is true of benzoylecgonine for cocaine, and of the glucuronide that dominates a lorazepam result.

NHTSA says this itself. Its 2024 fact sheets state that detecting THC metabolites in urine “only indicates prior exposure and cannot be correlated to drug impairment,” and that those urine detection times “are longer than windows of intoxication and impairment, even in occasional cannabis users.” That is the federal highway safety agency, in the document written for the prosecutors and drug recognition experts who work these cases.

Drug by drug

Ranges, not predictions. Read the source column: these come from controlled human dosing studies and manufacturer labels, and where studies disagree this page says so rather than picking the convenient one.

Most of the figures above come from NHTSA’s own 2024 fact sheets, the reference its drug recognition experts and traffic safety prosecutors are trained on. Where a row cites a study or a manufacturer’s label instead, that is because NHTSA does not cover that drug. Butalbital, secobarbital and drug-specific benzodiazepine windows are still absent, because no source for them was verified. An empty row is better than a confident guess.

Cannabis, where the numbers genuinely fight each other

Cannabis is the drug most people are asking about, and it is the one where the published research openly disagrees. Both of these are real findings from real studies:

The long number. A 1985 study of chronic daily users under supervised abstinence found urine positives for as many as 46 consecutive days, and up to 77 days before a subject produced ten consecutive negatives. The mean was 27 days. Excretion was not a clean downward curve either: subjects went negative, then positive again, without using anything.

The short number. A 2015 study of chronic users, using a modern confirmatory method, found that 73 percent fell below the cutoff within two weeks.

These are not really contradictory. They used different cutoffs, different assays and different analytical methods, and that is the entire point: the number is a property of the test, not of the person. Any page that quotes only the 77 days is selling you something, and so is any page that quotes only the two weeks.

What is not in dispute is that a chronic user can be positive long after any possible effect has ended. In monitored abstinence, where by definition nobody was impaired, chronic daily users were still blood-positive for THC at 30 days. There is also published work on distinguishing new use from residual excretion, which matters directly if a second positive is being treated as proof of a second use.

What the cutoff does to the answer

Federal workplace testing cutoffs, in nanograms per milliliter, from the current authorized panels effective July 2025. These are not DUI cutoffs and not Florida laboratory cutoffs, and that distinction is exactly the kind of thing that gets glossed over:

Urine analyte	Screening	Confirmation
Marijuana metabolite	50	15
Cocaine metabolite	150	100
Morphine	2,000	4,000
6-acetylmorphine	10	10
Amphetamine and methamphetamine	500	250
Oxycodone and oxymorphone	100	100
Fentanyl	1	1

Cannabis makes the point cleanly. After a single dose, at the 50 nanogram screening cutoff, the detection window is under two days. Drop the cutoff to 20 nanograms and the same dose is findable for up to six. Nothing about the person changed.

Screening tests also miss things. In one study of urines confirmed positive for benzodiazepines, the common immunoassay platforms detected only 47 to 78 percent of them, because they cross-react poorly with the metabolites that actually appear in urine. In another, patients genuinely prescribed clonazepam screened positive 21 percent of the time by immunoassay but 87 percent by the confirmatory method. A screen is not a result.

Does a blood THC number prove impairment

This deserves an honest answer rather than a convenient one, because the research cuts both ways and any lawyer telling you otherwise has not read it.

Against. A 2021 meta-regression covering 28 publications and 822 driving-related outcomes found blood THC correlated with impairment at roughly minus 0.08 to minus 0.10, which is very weak, and concluded that blood and oral fluid THC are relatively poor indicators of impairment. Critically, it found no significant relationship at all in regular users. Separate work found that the same dose impaired occasional users on several measures while leaving heavy users largely unaffected except on one task.

For. A driving simulator study found that blood THC of 8.2 and 13.1 micrograms per liter increased lane weaving about as much as breath alcohol of 0.05 and 0.08. A 2013 review concluded that recent smoking and blood THC of 2 to 5 nanograms per milliliter are associated with substantial impairment, particularly in occasional smokers. An international expert group proposed that serum THC of 7 to 10 nanograms corresponds to impairment comparable to a 0.05 blood alcohol level, while calling its own limit preliminary.

The defensible reading is narrower than either camp’s slogan. A blood THC concentration carries some information about impairment in an occasional user, and close to none in a regular user, where the drug persists for days or weeks and tolerance blunts the effect. No published threshold has been shown to separate impaired from unimpaired drivers across both groups. That is a stronger position than claiming the number means nothing, and it has the advantage of being true.

Crash risk is likewise more modest than commonly asserted. A 2016 reanalysis correcting two earlier meta-analyses put the pooled odds ratio at about 1.36, with a confidence interval of roughly 1.15 to 1.61, and reported that both previously published estimates fell outside that interval. For context, that is a modest increase next to alcohol. The paper carries a 2018 correction and drew several formal comment letters, and the reviews published since have continued to cite the same figure.

What this means in a Florida case

Florida has no per se limit for drugs the way it has 0.08 for alcohol. The State has to prove impairment, and a positive test is not impairment. That gap is where these cases are actually fought.

So the questions that matter are which fluid was taken and when, which molecule the laboratory reported, whether it is psychoactive at all, what cutoff was used, and whether anything in the record connects the concentration to how the person was driving. Parent drug versus metabolite covers the analyte question, blood versus urine covers the matrix, and tolerance covers why the same number means different things in different people.

One more thing that follows from the table above: a drug result cannot be run backwards in time the way the State tries to do with alcohol. Retrograde extrapolation is difficult enough with ethanol. With drugs there is no equivalent, and the State’s own experts will usually concede it.

Common questions

How long does THC stay in your system?

There is no single answer, and anyone who gives you one is guessing. It depends on how much, how often, which body fluid is tested and above all what cutoff the laboratory used. A single dose at the federal urine screening cutoff is detectable for under two days. Chronic daily use has been measured in weeks, and the published studies openly disagree about how many. The number is a property of the test, not of you.

Does a positive drug test mean I was impaired when I was driving?

No, and the government says so. NHTSA’s 2024 fact sheets state that detecting THC metabolites in urine only indicates prior exposure and cannot be correlated to drug impairment, and that urine detection times are longer than the windows of intoxication and impairment even in occasional users. For several drugs the molecule the laboratory looks for is not the one that affects you at all.

Can a drug test show when I used?

Rarely, and not in the way people assume. Most tests find a drug or its metabolite and say nothing about timing. In chronic users, published work has had to develop methods to tell genuinely new use apart from residual excretion, which matters directly if a second positive is being treated as proof of a second use. A few analytes are useful markers of recency, but they are the exception.

Does Florida have a legal limit for drugs, like 0.08 for alcohol?

No. Florida has no per se limit for drugs. For alcohol the State can prove the case with a number alone. For drugs it has to prove your normal faculties were impaired, and a laboratory result is not impairment. That gap is where these cases are fought.

Can a drug result be worked backwards to the time of driving?

There is no reliable equivalent of retrograde extrapolation for drugs. That calculation is difficult enough with alcohol, which has relatively predictable elimination. Drugs do not behave that way, the variation between people is enormous, and the State’s own experts will usually concede the point.

Where these figures come from

Every source is linked. Federal cutoffs come from the current authorized testing panels; pharmacokinetics come mainly from NHTSA’s 2024 fact sheets, with peer-reviewed controlled dosing studies where NHTSA is silent, and manufacturer labeling for the few drugs it does not cover. Each link was checked before publication, and where a paper has since been corrected the correction is linked beside it.

Where a figure could not be traced to a source it was left out rather than estimated. If you are a lawyer and want this with pinpoint citations, it is in The Florida DUI Trial Manual.