Evidence derived from in vitro experiments would suggest that cocaine exposure may hasten the progression of HIV disease among infected individuals. Epidemiologic support for this association is equivocal at best. We examined the relationship between cocaine use and decline in CD4 cell counts over a 6-month period in a cohort of 81 heterosexually active men and women who were infected with HIV. Overall. cocaine users were 1.4 times (90% CI = 1 .O-2.1) more likely to experience a decline in CD4 count than were non-cocaine users. Cocaine users with a baseline CD4 count of greater than 500 cells/mm' were at 1.6 times (90% C1= 1.2-2.3) greater risk for a CD4 decline than non-cocaine users at this baseline CD4 level. Concurrent treatment with an antiretroviral agent [AZT] modified the strength of this association, as evidenced by a cumulative incidence ratio (CIR) of 0.4 (90% CI =0.1-1.3) among AZT users and a CIR = 2.2 (go0/, CI = 1.5-3.2) among those not undergoing AZT treatment. The results of this study raise concerns about the negative effects of cocaine on people living with HIV infection, particularly those not receiving antiretroviral therapy who entered our study with a relatively intact immune system.