26 patients with systemic lupus erythematosus (SLE) induced by treatment with the antihypertensive drug hydralazine were investigated to determine if predisposition to the toxic effect was associated with an HLA-DR antigen. 25 of the 26 patients were slow acetylators. The group was compared with three others (1) 113 healthy subjects, untested for acetylator phenotype, (2) 16 slow-acetylator hypertensive patients treated with hydralazine for more than a year without developing SLE, and (3) 20 patients with idiopathic SLE. The frequency of HLA-DR4 (73%) was significantly higher in the group with hydralazine-induced SLE than in the other groups (respectively 33%, 25%, and 25%). The ratio of women to men affected was 4:1. If the slow acetylators treated with hydralazine were analysed as one group, it was observed that all women with DR4 developed hydralazine-induced SLE; the only men to do so were those with DR2 who were receiving 200 mg hydralazine per day. These observations have led us to suggest guide lines for hydralazine therapy and point to a striking association between an HLA-DR antigen and an adverse reaction to a therapeutic agent. It was also noted that the distribution of DR antigens in the hydralazine-SLE patients was significantly different from that in the group with idiopathic SLE. This supports the view that the syndromes are separate entities.