Solid-phase microextraction (SPME) is applied to the determination of cannabidiol, delta 8-tetrahydrocannabinol (delta 8-THC), delta 9-tetrahydrocannabinol (delta 9-THC), and cannabinol in pure water and human saliva. The inherent extraction behavior of the cannabinoids in pure water is evaluated along with optimization of the method in human saliva. The commercially available poly(dimethylsiloxane) (PDMS) SPME fibers were found to be the best class for the cannabinoid analysis. Partition coefficients were found to be extremely large for all of the cannabinoids (log K > 4.0). Equilibrium times for the 7- and 30-micron PDMS fibers were 50 and 240 min, respectively. A shorter extraction time of 10 min with the 30-micron PDMS fiber may be used for multiple extractions from the same vial, thus conserving the sample necessary for analysis and speeding up the total analysis time. Recoveries for the cannabinoids in saliva, relative to pure water, were dramatically improved by a method developed in our laboratory involving addition of glacial acetic acid to the sample vial prior to performing SPME. Using this method, recoveries relative to SPME in pure water ranged from 21 to 47% depending on the cannabinoid. The linear range for spiked saliva samples was established at 5-500 ng/mL (r2 > 0.994) with precisions between 11 and 20% RSD. The ultimate level of detection by SPME for the cannabinoids in saliva was 1.0 ng/mL, with signal-to-noise values of > or = 12. A saliva sample collected 30 min after marijuana smoking was subject to SPME and traditional liquid-liquid extraction analysis. Internal standard quantitation results for delta 9-THC by both methods yielded comparable results, indicating that the SPME method of analysis is highly accurate and precise. The level of delta 9-THC by SPME was found to be 9.54 ng/mL for the saliva sample.