Buprenorphine is a potent opioid analgesic used in the treatment of moderate to severe pain. At higher doses, it has demonstrated potential for treating heroin dependence. This study was undertaken to investigate buprenorphine pharmacokinetics by different routes of administration at dosages approximating those used in opioid-dependence studies. Six healthy men who were nondependent but who had a history of heroin use were administered buprenorphine in a crossover design study by intravenous (1.2 mg), sublingual (4.0 mg), and buccal (4.0 mg) routes of administration. Plasma samples were collected up to 96 h and assayed for buprenorphine and norbuprenorphine by negative chemical ionization tandem mass spectrometry. Plasma concentrations of buprenorphine and norbuprenorphine were analyzed by nonlinear regression analysis with standard noncompartmental methods. Buprenorphine biovailability by the sublingual and buccal routes was estimated as 51.4% and 27.8%, respectively, although there was considerable interindividual variability by both routes of administration. The terminal elimination half-lives were longer for the sublingual and buccal routes than for the intravenous route. The extended elimination half-lives may be due to a shallow depot effect involving sequestration of buprenorphine in the oral mucosa. Norbuprenorphine mean peak plasma concentrations were less than 1 ng/mL and were highly variable among different routes of administration and individuals. The terminal elimination half-life of norbuprenorphine was longer than buprenorphine.