1. To determine whether dexfenfluramine is a substrate of cytochrome P450 2D6 (CYP2D6), its disposition has been studied in nine extensive (EM) and eight poor metabolizers (PM) of debrisoquine. 2. Following a 30 mg dose of dexfenfluramine hydrochloride, urine was collected in all subjects for 96 h post-dose and plasma samples were collected in 11 subjects (six EMs and five PMs). Dexfenfluramine and nordexfenfluramine were measured in urine by h.p.l.c. and in plasma by g.c. 3. Urinary recovery of dexfenfluramine was greater in PMs than EMs (4136 +/- 1509 micrograms vs 1986 +/- 792 micrograms; 95% CI of difference 926-3374; P < 0.05) whereas that of nordexfenfluramine was similar in both phenotypes (PM: 1753 +/- 411 micrograms vs 1626 +/- 444 micrograms). 4. Dexfenfluramine AUC was higher in PMs (677 +/- 348 micrograms l-1 h) than EMs 359 +/- 250 micrograms l-1 h). The apparent oral clearance of dexfenfluramine was greater in EMs than PMs (93.6 +/- 42.4 l h-1 vs 45.6 +/- 19.5 l h-1; 95% CI of difference 1.2-94.7; P < 0.05). The renal clearance was similar in both phenotypes (EMs: 5.88 +/- 2.83 l h-1; PMs 6.60 +/- 2.01 l h-1), indicating that the higher urinary recovery of dexfenfluramine in PMs reflects higher plasma concentrations, rather than phenotype differences in the renal handling, of dexfenfluramine. 5. The apparent nonrenal clearance of dexfenfluramine was substantially lower (P < 0.05; 95% CI of difference 3.0-94.1) in PMs (39.0 +/- 19.5 l h-1) than EMs (87.6 +/- 41.2 l h-1). 6. There was a significant inverse correlation (rs = 0.776 95% CI-0.31-0.94; n = 11; p = 0.005) between the debrisoquine metabolic ratio and the apparent nonrenal clearance of dexfenfluramine. 7. PMs had a higher incidence of adverse effects (nausea and vomiting) than EMs. 8. In conclusion, the metabolism of dexfenfluramine is impaired in PMs. Thus CYP2D6, the isoenzyme deficient in poor metabolizers of debrisoquine, must catalyse at least one pathway of dexfenfluramine biotransformation.