Oral fluid (OF) testing offers noninvasive sample collection for on-site drug testing; however, to date, test performance for Δ(9)-tetrahydrocannabinol (THC) detection has had unacceptable diagnostic sensitivity. On-site tests must accurately identify cannabis exposure because this drug accounts for the highest prevalence in workplace drug testing and driving under the influence of drugs (DUID) programs. Ten cannabis smokers (9 males, 1 female) provided written informed consent to participate in this institutional review board-approved study and smoked 1 6.8%-THC cigarette ad libitum. OF was collected with the Draeger DrugTest(®) 5000 test cassette and Quantisal™ device 0.5 h before and up to 22 h after smoking. Test cassettes were analyzed within 15 min (n = 66), and Quantisal GC-MS THC results obtained within 24 h. Final THC detection times and test performances were assessed at different cannabinoid cutoffs. Diagnostic sensitivity, diagnostic specificity, and efficiency at DrugTest 5000's 5 μg/L screening cutoff and various THC confirmation cutoffs were 86.2-90.7, 75.0-77.8, and 84.8-87.9%, respectively. Last detection times were >22 h, longer than previously suggested. Confirmation of 11-nor-9-carboxy-THC, absent in THC smoke, minimized the potential for passive OF contamination and still provided 22-h windows of detection, appropriate for workplace drug testing, whereas confirmation of cannabidiol, and/or cannabinol yielded shorter 6-h windows of detection, appropriate for DUID OF testing. The DrugTest 5000 on-site device provided high diagnostic sensitivity for detection of cannabinoid exposure, and the selection of OF confirmation analytes and cutoffs provided appropriate windows of detection to meet the goals of different drug testing programs.