Can Whole-Blood THC Exceed 5 ng/mL Days After Cannabis Use Without Worse Simulator Driving? A 190-Person Study

What question did the original researchers test?

The authors asked two linked questions. First, how frequently do active, regular cannabis users still have quantifiable parent THC in their whole blood after being instructed not to use cannabis for at least two days? Second, among those same users, are the baseline driving-simulator scores different when a measured THC result is above rather than below selected numerical thresholds used in some cannabis-driving statutes? Those questions are narrower than whether a particular jurisdiction’s law is justified and far narrower than whether a specific crash was caused by cannabis.

A per se concentration threshold is an administrative or legal decision rule. An analytical measurement records the amount of a specified chemical in a specified specimen at a particular collection time. Driving impairment is a functional and contextual question. A chemical cutoff can be compared with actual performance, but the comparison must not quietly redefine one as the other. This study supplied a direct comparison at baseline; it did not validate the thresholds as individual impairment classifiers.

The researchers also followed whole-blood THC after an experimental smoked-cannabis or placebo session. That later series provided information about the return of concentrations toward each person’s baseline. It did not transform the baseline comparison into an on-road driving test or an individualized time-since-use formula.

Who was studied, and was abstinence independently verified?

The study analyzed 190 distinct adults who regularly used cannabis. Eligibility required use at least four times in the preceding month, age 21–55 years, a valid driver’s license, and at least 1,000 miles driven in the previous year. These were not 190 roadside arrests, not randomly sampled drivers, and not 190 people newly exposed to a standard dose of cannabis. The findings therefore describe this selected research population, not the prevalence of residual THC in every driver.

Participants were instructed to abstain for at least 48 hours before baseline phlebotomy. The self-reported median time since last cannabis use was three days, with an interquartile range of three to six days. An oral-fluid THC screen was used to exclude people whose result exceeded 5 ng/mL. Importantly, the team did not keep participants in a secure abstinence unit for the preceding two days. Self-report and the oral-fluid exclusion criterion help assess recent use, but neither independently proves each individual’s last consumption time.

This distinction matters in forensic phrasing. It is accurate to say that a blood result above 5 ng/mL occurred after a requested abstinence interval, in participants who generally reported several days without use. It would overstate the source to say that every blood sample was proven to follow 48 hours of continuous observed abstinence.

What exactly did the laboratory measure?

The investigators quantified THC and related cannabinoids in directly collected whole blood by isotope-dilution liquid chromatography–tandem mass spectrometry (LC–MS/MS). For parent THC, the method’s lower limit of quantification was 0.5 ng/mL. The authors describe approximately ±20% accuracy at this lower quantification limit. A number below that limit is analytically distinct from proof of absolutely zero THC in circulation.

Whole blood means the unseparated blood specimen used in the study, not centrifuged plasma or clinical serum. THC partitions between blood components; no validated universal conversion from a plasma value to a whole-blood value follows from this paper. Neither urinary THC-COOH nor an oral-fluid screening result can be silently substituted for the study’s parent whole-blood THC concentration. The separate guide to biological matrices explains why specimen identity must stay attached to each reported number.

Other compounds were quantified in the same blood work, including 11-hydroxy-THC, the carboxy metabolite THC-COOH, and THC-COOH glucuronide. The authors did not hydrolyze THC-COOH glucuronide into free THC-COOH for this assay; those reported entities therefore remained analytically distinct. A metabolite value is not a second measurement of parent THC and cannot be treated as an independent indicator of actual present driving performance.

How many baseline samples were above 0.5, 2, and 5 ng/mL?

Of 190 study participants, 43% were reported at or above 0.5 ng/mL, 45 of 190 (about 24%) were at or above 2 ng/mL, and 10 of 190 (5.3%) were at or above 5 ng/mL. The 0.5 ng/mL threshold reflected the paper’s quantification limit and its operational comparison with so-called zero-tolerance policies; it is not a universal statutory definition of zero tolerance. Jurisdiction-specific laws, approved specimen types, and operative dates require separate legal verification.

The comparison of driving performance used 187 rather than all 190 participants, so the threshold group counts in the performance figure differ slightly from those in the complete blood-chemistry cohort. The paper’s Figure 3 provides the actual driving-group denominators. A denominator is not a minor technicality: the blood-prevalence question and the driving-score question are separate analyses of different available records.

These are the original thresholds and unadjusted baseline comparisons reported by the authors, not newly selected statistical cutoffs. Their failure to produce significant group-score differences is not an equivalence test, a prediction-accuracy estimate, or a demonstration that no participant experienced cannabis-related effects. It shows that dividing this cohort by these measured blood concentrations did not significantly separate its baseline composite simulator scores.

How much parent THC and metabolite remained in whole blood?

The study’s baseline mean THC was 1.1 ng/mL (standard deviation 2.0), but its median was reported as 0 because many results were not quantifiable above the 0.5-ng/mL threshold. The maximum was 16.2 ng/mL. This distribution is skewed: quoting the mean without the median, or the maximum without the sample size, would misrepresent a substantial part of the measured experience.

The metabolite results illustrate another problem with collapsing an entire cannabinoid profile into the word “THC.” THC-COOH is a downstream carboxy metabolite; THC-COOH glucuronide is its conjugate. Their blood values can be much higher than those of parent THC. They were determined in the same direct whole-blood matrix, yet neither replaces the parent result in a THC-specific concentration rule.

Zero in a summary table partly reflects values under an instrument’s quantification limit. It should not be translated into a population claim that all underlying physical samples had no molecules of the analyte. Conversely, very high carboxy-metabolite values do not allow the analyst to calculate a particular person’s time since last cannabis use without a validated case-specific kinetic basis.

Did higher blood THC distinguish baseline driving performance?

Baseline driving was measured using a 25-minute simulated drive. The composite score summarized performance variables including variation in lateral position, speed variability, car following, and correct divided-attention responses. It was not a direct count of real-world crashes, and participants were not tested on an open public road as part of this analysis.

For the 5-ng/mL split, 10 people at or above the threshold were compared with 177 below it, with p = 0.14. For 2 ng/mL, the split was 44 versus 143, p = 0.36. For the 0.5-ng/mL quantification comparison, 79 were at or above and 108 below, p = 0.40. None reached the conventional two-sided 0.05 criterion. Age, sex, and body mass index differences did not explain the associations with the tested thresholds in the supplementary analyses described by the authors.

A non-significant p value is not evidence that the groups were perfectly equivalent. Small high-concentration subgroups can leave substantial uncertainty, especially the 10-person group at 5 ng/mL. Nor does the result reveal what any specific participant’s score would have been had that person never used cannabis. This study lacked a matched group of drivers who did not use cannabis, a limitation the researchers themselves identified.

For legal-scientific interpretation the appropriate conclusion is narrower: these numerical parent whole-blood THC cutpoints did not discriminate the selected baseline simulator-score groups in this research sample. Whether a specific driver showed unsafe attention, roadway errors, poor judgment, or drug-related effects requires independent evidence rather than a binary group comparison.

How did smoking affect blood THC five hours later?

After baseline collection, study participants entered experimental active-cannabis or placebo smoking conditions; the THC-content arms included 5.9% and 13.4% cannabis material. The reported post-smoking change in whole-blood THC was calculated by subtracting the baseline value from the final draw approximately five hours after starting to smoke. This is a within-person chemical difference, not a standard impairment score.

Among 104 active-drug participants with the reported comparison, the median final-minus-baseline difference was 0.5 ng/mL, interquartile range approximately 0–0.9 ng/mL, and the one-sample comparison with zero was significant (p < 0.001). Among 51 placebo participants, the median change was 0 (0–0) ng/mL; the mean-change test was not significant, p = 0.69.

The paper reports 38 active-arm participants with a change of zero or less and describes roughly 37% as having returned to baseline at five hours. Those details help explain how quickly a post-smoking blood concentration can move toward a regular user’s existing residual level. They do not show that the active THC dose had no earlier physiological or functional effects, nor does a near-baseline final value prove the person had not smoked that day.

The whole-blood difference is particularly important when reconstructing sequences of samples. Without a known prior baseline, a late low numerical concentration can be consistent with different routes, amounts, and time courses. A standard pharmacokinetic interpretation must preserve uncertainty about dosing, individual elimination, and the timing of collection.

Why was the participant with 16.2 ng/mL informative but not decisive?

The highest pre-smoking parent whole-blood THC concentration was 16.2 ng/mL. The authors examined that participant’s subsequent concentration-time profile because such a value could raise an obvious question about very recent use. At 14 minutes after smoking 580 mg of 5.9%-THC cannabis, the measured concentration reached 135 ng/mL; by roughly 4 hours 55 minutes it was 16.5 ng/mL.

Her baseline blood also contained 11-hydroxy-THC at 9.3 ng/mL, free THC-COOH at 133 ng/mL, and THC-COOH glucuronide at 571 ng/mL. She described use on 27 of the previous 30 days and reported approximately two and a half days without cannabis before the visit. Her pre-smoking composite simulator performance was described as slightly better than the group’s average.

The return toward the unusual initial value is consistent with a persistent residual concentration in a frequent user. But one individual’s trajectory cannot retrospectively prove an exact abstinence interval, validate her reporting, or establish that every driver at 16.2 ng/mL has normal driving ability. The method’s uncertainty near lower concentrations and the absence of sequestered abstinence add further reasons not to turn the case into a universal rule.

Is a whole-blood concentration a timestamp for cannabis use?

No. A single concentration is a chemical measurement at a known sampling time, not a clock. After inhalation, parent THC can increase rapidly, redistribute, and decline while residual concentrations remain in some regular users. The original study’s five-hour sampling comparison and the high-baseline individual demonstrate that a similar observed number may appear at different parts of an exposure history.

Potential reconstruction also depends on whether a person smoked, vaporized, swallowed an edible, or used an oromucosal product. A dose printed on a package or expressed as percentage potency is not an independently measured absorbed dose. This study involved smoked cannabis under research conditions; it did not measure edible dosing, individual tissue stores, or a universally validated back-extrapolation rate for human THC.

It is therefore not defensible to subtract a population average from an isolated blood result and announce a precise hour of last use. A record-bound forensic reconstruction would examine collection times, chain of custody, analytical methods, complete serial concentrations when available, statements, and independent behavior—not substitute an attractive but unsupported pharmacokinetic formula.

What does a positive THC result establish, and what does it leave unanswered?

A properly identified and quantified whole-blood parent THC result establishes that the specified analyte was measured in the submitted specimen above the method’s reporting criteria. It can help establish exposure and may contribute to a broader evaluation. It does not alone answer route, exact consumed quantity, precisely when use ended, driving competence, legal culpability, or why a roadway event happened.

The analytical chain matters. Which vessel was collected? Was the specimen whole blood, serum, or plasma? Did the laboratory measure the parent compound using adequate calibrators, blanks, controls, retention-time rules, and ion transitions? Was the numerical result within its validated quantification range? The source paper used quantitative LC–MS/MS; the contextual confirmatory drug-testing guide explains why independent laboratory validation remains important in a case record.

Once the chemical result is reliable, effect remains a separate evidentiary question. Witness observations, road circumstances, psychomotor tasks, medical findings, competing causes, and collection delays may all matter. Some observations can be nonspecific, and any apparent evidence of impairment must be evaluated against relevant alternative explanations. The present study did not test a complete forensic assessment algorithm that combined these features.

Why should prosecutors and defense experts both preserve the study’s limits?

For a prosecutor, a quantifiable concentration may have statutory significance, and the actual law may define a threshold without requiring the same scientific proof of individual driving deterioration as a clinical impairment test. This research article cannot repeal a statute or determine admissibility, jury instructions, or liability in a particular jurisdiction. Those are questions for current governing law and the actual case record.

For a defense expert, the paper documents a measurable false assumption that can enter toxicological argument: a specific parent whole-blood THC number does not automatically identify a recent-use time or reliably separate simulator driving scores in every regular user. But the researcher should not convert a group-level lack of statistical significance into an assertion that a particular defendant was unimpaired. Neither side benefits scientifically from making the paper prove more than it measured.

The critical separation is between statutory classification, analytical quantification, observed functional evidence, and individual event causation. Different types of evidence may inform different elements; they are not synonyms, and their legal significance depends on the relevant jurisdiction and instructions.

Could the results reflect tolerance or frequent-use history?

Possibly, but this experiment does not isolate tolerance as the causal explanation for the null threshold comparisons. The study population consisted of regular users, and higher baseline THC groups reported greater use frequency, more recent use, and larger consumption. Those patterns may reflect accumulation, individual pharmacokinetics, recent behavior, or other characteristics associated with frequent use. The paper did not randomize participants to become tolerant or measure a unique tolerance coefficient.

Moreover, matching higher and lower whole-blood THC groups on a handful of demographics is not the same as controlling every behavioral or pharmacologic confounder. The absence of a cannabis-naive comparison group means the analysis tests how subgroups within regular users differed, rather than whether regular users resemble abstinent nonusers in simulator performance.

It would be scientifically improper to assert that habitual exposure immunizes anyone against THC effects. Other research might address tolerance, but the present original paper’s key empirical point is the limited discriminatory performance of selected concentration cutpoints among its particular baseline cohort.

Which limitations most affect forensic generalization?

First, the abstinence interval was requested, not continuously observed. The investigators relied on reported history and an oral-fluid screening exclusion; a missed recent use episode cannot be ruled out by study design alone. Second, the participants were a convenience research cohort of regular users meeting license and driving-experience requirements, not a representative sample of all drivers.

Third, simulator outcomes do not capture every hazard on a public road. Their composite combines several performance metrics and may not detect every form of impairment or individual error. The 187-person driving subset excludes three of the 190 baseline chemistry records for reasons not fully resolved within this paper. Fourth, there is no matched never-user control, no direct proof of zero impairment, and no reported sensitivity, specificity, or receiver-operating-characteristic threshold performance that would validate an individual diagnostic use of the cutpoints.

Fifth, some subgroups were small, especially 10 people above 5 ng/mL in the driving comparison. A nonsignificant p value can arise from limited precision as well as genuinely small differences. The source should not be used to assert equivalence margins or narrow confidence intervals that it did not establish. Finally, the paper evaluated numerical comparisons selected in its 2025 legal-scientific context, not an up-to-date inventory of laws in every jurisdiction in 2026 or later.

What information belongs in a case-specific scientific review?

A meaningful review begins with the charging and collection timeline, not merely the final number. Retrieve the original laboratory report and supporting chromatograms, specimen collection and storage records, LC–MS/MS identification criteria, calibrator and quality-control performance, measurement uncertainty, and any earlier or later blood draws. Note whether the reported value is parent THC, 11-hydroxy-THC, THC-COOH, or a conjugate. Do not relabel serum as whole blood or assume a breath or urine result supplies the same chemical fact.

Separately examine the reported use history, any independently documented administration, contemporaneous observations, road conditions, medical factors, and recorded functional tests. A particular test score may be affected by practice, fatigue, distraction, vision, disease, or other drugs. These facts require disciplined evaluation of competing hypotheses rather than assigning an individual probability of impairment from a population summary.

Finally, document the applicable legal question separately from the scientific one. A numerical threshold may have a statutory consequence independent of whether a paper shows group-score separation. Equally, the absence of a threshold violation does not automatically establish safe operation. An expert can state what the study supports, acknowledge contrary evidence, and identify precisely what information remains missing.

Does this paper prove that cannabis cannot impair driving?

No. This was not the hypothesis it tested, and that conclusion would directly confuse residual analytical detection with time-dependent drug effects. The researchers note that earlier work using related experimental methods found diminished driving performance during acute post-smoking intervals. The fact that a person may have measurable THC after effects have diminished does not mean effects never occur; likewise, a statistically unremarkable baseline group score does not prove an individual performed safely.

The paper’s strongest original contribution is that directly measured parent whole-blood THC can remain at or above selected numerical cutpoints in people reporting multi-day abstinence, while those cutpoints failed to distinguish the study’s contemporaneous baseline simulator scores. It supports skepticism about treating a lone concentration as a universal personal impairment classifier, not skepticism about the possibility of impairment itself.

Evidence-based conclusion

Among 190 selected regular cannabis users asked to abstain for at least 48 hours, the original 2025 study observed 45 whole-blood parent-THC results at or above 2 ng/mL and 10 at or above 5 ng/mL, with a maximum 16.2 ng/mL. Across 187 participants with baseline simulator data, these and the 0.5-ng/mL cutoff did not distinguish group composite scores at conventional statistical significance. Serial post-smoking results returned near the pre-smoking baseline by about five hours for many participants, yet that pattern is not a universal pharmacokinetic clock.

The defensible forensic answer is therefore conditional: a measured THC concentration can outlast the most obvious acute-use window and cannot, by itself, certify either impairment or safety for an individual person. Use specimen-specific confirmatory chemistry and independently evaluated functional and incident evidence. Do not transform a participant-level research association, a null hypothesis test, or a jurisdiction’s statutory label into a scientific conclusion the source did not establish.

Frequently asked questions

1. Does 5 ng/mL whole-blood THC prove a driver was impaired?

No. Ten participants in the original chemistry cohort had at least 5 ng/mL after requested abstinence, and the baseline simulator group comparison above versus below this level was not statistically significant (p = 0.14, 10 versus 177). That result does not show that every such driver is unimpaired; legal consequences depend on current jurisdiction-specific law and the case record.

2. Was cannabis abstinence for two days verified in a secure facility?

No. Participants were instructed to abstain for at least 48 hours and self-reported a median three-day interval. Oral-fluid screening excluded individuals above the study’s THC screening criterion. They were not continuously confined or monitored for the preceding two days, so precise last-use timing was not independently established for everyone.

3. Why was the study median THC reported as zero if some users had high concentrations?

Many results fell below the assay’s 0.5-ng/mL quantification limit, producing a median of zero in its reported summary, even though its measured range extended to 16.2 ng/mL. Below-quantification is not proof that a biological specimen contained absolutely no THC.

4. Do the findings apply to serum, plasma, oral fluid, or urine THC tests?

Not as direct numeric equivalences. The primary cutpoint analysis used directly assayed whole-blood parent THC. Serum/plasma distribution, urinary metabolite excretion and oral-fluid testing represent different matrices, analytes or detection windows and require their own method validation and interpretive evidence.

5. Could a THC concentration five hours after smoking look similar to the person’s pre-smoking concentration?

Yes. The active-arm median whole-blood change from baseline at about five hours was 0.5 ng/mL, and a substantial fraction returned to their original baseline. This does not mean no acute effects occurred earlier, nor does it timestamp a particular person’s last smoking episode.

6. Does a nonsignificant simulator p value establish that the two groups are identical?

No. The paper found p = 0.14, 0.36 and 0.40 for its selected cutpoint comparisons; none is an equivalence study. Differences may exist that this design and sample size did not reliably detect. The simulator also did not reproduce every relevant aspect of real-world driving.

7. Can an expert use this study to reconstruct a defendant’s exact last cannabis use?

No. The study did not validate an individual backward-calculation algorithm from one parent whole-blood THC value. Such reconstruction would require case-specific analytical, temporal, exposure and clinical evidence, with uncertainty stated explicitly; a numerical concentration alone is insufficient.