Oral fluid (OF) testing is increasingly important for drug treatment, workplace, and drugged-driving programs. There is interest in predicting plasma or whole-blood concentrations from OF concentrations; however, the relationship between these matrices is incompletely characterized because of few controlled drug-administration studies. Ten male daily cannabis smokers received around-the-clock escalating 20-mg oral Δ(9)-tetrahydrocannabinol (THC, dronabinol) doses (40-120 mg/day) for 8 days. Plasma and OF samples were simultaneously collected before, during, and after dosing. OF THC, 11-hydroxy-THC and 11-nor-9-carboxy-THC (THCCOOH) were quantified by GC-MS at 0.5-μg/L, 0.5-μg/L, and 7.5-ng/L limits of quantification (LOQs), respectively. In plasma, the LOQs were 0.25 μg/L for THC and THCCOOH, and 0.5 μg/L for 11-hydroxy-THC. Despite multiple oral THC administrations each day and increasing plasma THC concentrations, OF THC concentrations generally decreased over time, reflecting primarily previously self-administered smoked cannabis. The logarithms of the THC concentrations in oral fluid and plasma were not significantly correlated (r = -0.10; P = 0.065). The OF and plasma THCCOOH concentrations, albeit with 1000-fold higher concentrations in plasma, increased throughout dosing. The logarithms of OF and plasma THCCOOH concentrations were significantly correlated (r = 0.63; P < 0.001), although there was high interindividual variation. A high OF/plasma THC ratio and a high OF THC/THCCOOH ratio indicated recent cannabis smoking. OF monitoring does not reliably detect oral dronabinol intake. The time courses of THC and THCCOOH concentrations in plasma and OF were different after repeated oral THC doses, and high interindividual variation was observed. For these reasons, OF cannabinoid concentrations cannot predict concurrent plasma concentrations.