A retrospective cohort study was conducted to analyze variables predictive of acquired immunodeficiency syndrome (AIDS), as defined by the Centers for Disease Control (CDC) in 1987, and death in 224 women infected with the human immunodeficiency virus (HIV) who were evaluated through two HIV outpatient programs in New Orleans between January 1987 and December 1991 and followed through February 1993. Variables predictive of AIDS in a multivariate proportional hazards model were an entry CD4 cell count < 200/mm3 (adjusted RR 8.2, 95% CI 3.2, 20.9) and rapid CD4 cell count decline (adjusted RR 6.0, 95% CI 2.4, 15.5). A baseline AIDS diagnosis (by the CDC 1987 definition) was highly predictive of death (RR 36.3, 95% CI 15.7, 70.1). In multivariate analyses none of the relative risks for predictive variables remained significant if adjusted for a diagnosis of AIDS at entry. Although immunological parameters were predictive of early HIV disease progression, an AIDS diagnosis was the strongest prognostic indicator of death in a cohort of North American women.