A decline in the proportion of Kaposi's sarcoma among AIDS cases since the 1980s has been attributed to changes in sexual behavior among homosexual men and a decreasing exposure to a hypothesized Kaposi's sarcoma cofactor. Recent studies have shown that the incidence rate of Kaposi's sarcoma has remained relatively stable, which seems to argue against the hypothesis of a declining exposure to the putative cofactor. To examine this paradox, we evaluated the incidence of Kaposi's sarcoma, using Cox proportional hazard analyses, and performed a simulation to compare incidences of different AIDS outcomes among 407 homosexual men with documented dates of seroconversion. Our data show that men who seroconverted early in the epidemic did not progress faster to Kaposi's sarcoma than men who seroconverted more recently. A lower incidence rate of Kaposi's sarcoma would be expected among the latter group if exposure to the hypothesized cofactor is decreasing over time. The percentage of Kaposi's sarcoma among incident AIDS cases decreased over the years following seroconversion, but not over calendar time. This study demonstrates that the decline in the proportion of Kaposi's sarcoma among AIDS cases should not be interpreted as a decline in the incidence of Kaposi's sarcoma and that there is no evidence that a hypothesized Kaposi's sarcoma cofactor is declining over calendar time.