The second AIDS-defining condition diagnosed chronologically is referred to in this report as the secondary AIDS diagnosis. In this study, we examined survival following a secondary AIDS diagnosis using demographic and clinical factors known within 1 year before secondary AIDS diagnosis. In a prospective cohort of 2412 HIV-seropositive homosexual men observed in the Multicenter AIDS Cohort Study (MACS), 609 presented with a secondary AIDS diagnosis between January 1, 1988 and March 31, 1995. To analyze the data, we used survival analysis methods including the Kaplan-Meier product-limit estimator and extended Cox models that allow for nonproportional hazards. The median survival time after a secondary diagnosis was 10.3 months. Rapidity of progression from an initial AIDS diagnosis to a secondary diagnosis was not associated with survival. Drug treatment did not show a beneficial effect because of confounding by indication (i.e., selection bias) and limited efficacy on advanced disease of treatments available prior to 1995. However, a beneficial effect was captured by the use of calendar periods as a proxy measure for the relative exposure to drug treatments. Later calendar year of secondary diagnosis, secondary Kaposi's sarcoma, and higher CD4+ cell count were found to be significantly (p < .05) associated with longer survival time. However, secondary AIDS diagnosis was a significant factor only in the short term. Using secondary Pneumocystis carinii pneumonia as the reference diagnosis, the relative hazard of death 3 months after the time of secondary Kaposi's sarcoma diagnosis was 0.56 (95% confidence interval [CI] = 0.36-0.89) whereas the relative hazard after concurrently diagnosed multiple secondary illnesses was 2.06 (95% CI = 1.26-3.38). After approximately 1 year from the secondary diagnosis, the type of diagnosis was no longer significantly associated with survival.