To evaluate the quality of studies about drugs and food interactions with warfarin and their clinical relevance. MEDLINE and TOXLINE databases from 1966 to October 1993 using the Medical Subject Headings warfarin, drug interactions, and English only. All articles reporting original data on drug and food interactions with warfarin. Each report, rated independently by at least two investigators (using causality assessment), received a summary score indicating the level of assurance (level 1 = highly probable, level 2 = probable, level 3 = possible, and level 4 = doubtful) that a clinically important interaction had or had not occurred. Inter-rater agreement was assessed using a weighted κ statistic. Of 793 retrieved citations, 120 contained original reports on 186 interactions. The weighted κ statistic was 0.67, representing substantial agreement. Of 86 different drugs and foods appraised, 43 had level 1 evidence. Of these, 26 drugs and foods did interact with warfarin. Warfarin's anticoagulant effect was potentiated by 6 antibiotics (cotrimoxazole, erythromycin, fluconazole, isoniazid, metronidazole, and miconazole); 5 cardiac drugs (amiodarone, clofibrate, propafenone, propranolol, and sulfinpyrazone); phenylbutazone; piroxicam; alcohol (only with concomitant liver disease); cimetidine; and omeprazole. Three patients had a hemorrhage at the time of a potentiating interaction (caused by alcohol, isoniazid, and phenylbutazone). Warfarin's anticoagulant effect was inhibited by 3 antibiotics (griseofulvin, rifampin, and nafcillin); 3 drugs active on the central nervous system (barbiturates, carbamazepine, and chlordiazepoxide); cholestyramine; sucralfate; foods high in vitamin K; and large amounts of avocado. Many drugs and foods interact with warfarin, including antibiotics, drugs affecting the central nervous system, and cardiac medications. Many of these drug interactions increase warfarin's anticoagulant effect.