To define the differences in codeine pharmacodynamics in extensive (EMs) and poor (PMs) metabolizers of debrisoquin and to determine whether the inhibition of codeine's metabolism by quinidine produces phenotypically dependent pharmacodynamic changes, we studied 16 healthy nonsmoking males, 10 EMs and 6 PMs of debrisoquin. The subjects received in random double-blind fashion 120 mg of codeine plus placebo, 120 mg of codeine plus 100 mg of quinidine and 100 mg of quinidine plus placebo. Blood was obtained over 24 hr and urine was collected for 48 hr. Respiratory, psychomotor and pupillary effects of codeine were greater in the EMs than in the PMs (P < .01). Morphine and morphine metabolites were detectable only in plasma from EMs. Codeine metabolic clearance by O-demethylation was almost 200-fold greater in the EMs than in the PMs. After coadministration of quinidine, morphine and morphine metabolites were not detectable in the plasma of either phenotype and mean (+/- S.E.M.) O-demethylation clearance was reduced in the EMs from 162.7 +/- 36.6 to 17.0 +/- 5.0 ml/min (P < .003), but not in the PMs. The diminished production of morphine in the EMs was associated with significantly reduced respiratory, psychomotor and pupillary effects (P < .01). Thus, CYP2D6 mediated O-demethylation of codeine to morphine is central to its pharmacodynamic effects. Patients who lack CYP2D6 or whose CYP2D6 is inhibited would not be expected to benefit from codeine. Thus, phenotyping for CYP2D6 and the avoidance of CYP2D6 inhibitors is justified in patients with chronic path before initiating long-term therapy with analgesics whose in vivo activation is dependent on CYP2D6 activity (i.e., codeine, hydrocodone and oxycodone.