Although efficacy and toxicity of primaquine (PQ) depend on bioconversion, the process is poorly understood, even for carboxyprimaquine (CPQ), the major plasma metabolite. Earlier work to clarify drug metabolism showed that PQ could be converted quantitatively into CPQ, in vitro, with human erythroleukemic K562 cells or nonleukemic bone marrow supplemented with calf serum. We have now found--using systems with serum only, as well as with K562, bone marrow, and adult or embryonic liver cells--that the bioconversion of the side chain of PQ involves a branched pathway with at least three separate enzymes and two derivatives other than CPQ. An oxidase activity in serum converted PQ first into a novel side chain aldehyde (Y). Aldehyde dehydrogenase transformed PQ-aldehyde into CPQ in cell-free systems and in K562, bone marrow, and adult liver cells. Embryonic hepatocytes or bone marrow treated with 1,3-bis(2-chloroethyl)-1-nitrosourea did not produce CPQ; instead, they made a metabolite (Xc) that we could synthetize via PQ-aldehyde and identify as PQ-alcohol. PQ-alcohol replaced CPQ as the final product whenever alcohol-dehydrogenase prevailed over aldehyde dehydrogenase. These enzymes operated in intact cells and controlled the biotransformation of PQ absolutely. Unless both dehydrogenase were absent, inhibited, or deprived of coenzyme, potentially cytotoxic PQ-aldehyde intermediate did not accumulate. Some of the unique tissues schizonticidal and gametocidal effects of PQ may depend on the distribution pattern and relative activities of PQ oxidase, aldehyde dehydrogenase, and alcohol dehydrogenase in human subjects and in parasites.