1 A single oral dose of codeine (25 mg) was given to 132 healthy Swedish Caucasians who had previously been phenotyped with respect to debrisoquine hydroxylation. The 'metabolic ratios' (MR) in urine of codeine O-demethylation (codeine/(morphine (M) + morphine-3-and 6-glucuronides (M3G and M6G) + normorphine)), N-demethylation (codeine/(norcodeine (NC) + norcodeine glucuronide + normorphine (NM))) and glucuronidation (codeine/codeine-6-glucuronide (C6G)) were calculated following h.p.l.c. analysis of urine samples collected over 8 h. 2 There was a significant correlation between the log MR for debrisoquine hydroxylation and the log MR for codeine O-demethylation (r, = 0.77, P < 0.001). The poor debrisoquine hydroxylators had MRs of codeine O-demethylation between 8.3 and 55.1, while the values for extensive hydroxylators were between 0.4 and 5.5. 3 The poor debrisoquine hydroxylators excreted significantly less M, M3G, M6G and NM, while the urinary recovery of C6G and NC was significantly higher in these subjects compared to the extensive hydroxylators. 4 The MRs for glucuronidation and N-demethylation did not exhibit a bimodal distribution, and were not related to the MR of debrisoquine hydroxylation. 5 No associations were found between sex, body-weight, smoking habits, age, urine volume or urine pH and the O-demethylation of codeine. 6 The O-demethylation of codeine to form M appears to be under the same polymorphic genetic control as the 4-hydroxylation of debrisoquine.