Two intravenous doses of tetrahydrocannabinol (THC) (0.022 mg/kg and 0.044 mg/kg) were compared to intravenous diazepam (0.157 mg/kg) and to placebo (Ringer's lactate) as premedication for dental extraction in 10 healthy volunteers. Pain detection and tolerance thresholds were measured and psychiatric interviews were supplemented by Minnesota Multiphasic Personality Inventories (MMPI), the Zung Depression Scale (ZDS), Beck Depression Inventories (BDI), and the State-Trait Anxiety Inventory (STAI). Pain detection thresholds were altered unpredictably with high THC doses, but analgesia as indicated by pain tolerance was less than that after diazepam and placebo. In three subjects low-dose THC (0.022 mg/kg) was a better analgesic than placebo but not diazepam. Six subjects preferred placebo to low-dose THC as an analgesic; this group experienced increases in subjective surgical pain and were submissive, rigid, and less introspective with high State Anxiety and MMPI profiles that differed from subjects whose pain was not increased. STAI following THC presaged a poor analgesic response in this group.