In our ongoing cohort study of homosexual men, the ratio of new Kaposi's sarcoma (KS) cases to new opportunistic infections (OI) during the periods 1982-1985, 1986-1987, and 1988-1989 fell from 0.75 (9 KS: 12 OI) to 0.57 (12 KS:21 OI) to 0.27 (4 KS:15 OI), respectively. To examine factors associated with the development of KS as compared to OI, we compared antecedent risk factors in 25 KS cases and 48 OI "controls." In univariate analyses, several classical HIV risk factors including numbers of sexual partners and receptive anal intercourse were higher in the KS than the OI group. The strongest associations were found with an elevated number of sex partners in high-risk areas (San Francisco, Los Angeles, and New York) in the 5 years prior to enrollment and with elevated use of nitrite inhalants. Logistic regression revealed the latter two variables and an elevated number of partners contacted in washrooms/parks to be significant, independent risk factors for KS relative to OI. Any or all of these variables could be related with early HIV infection. However, the association with early sexual contact in high-risk areas raises the more intriguing possibility that this variable is an indicator of an increased exposure either to a particular strain of HIV that is more pathogenic for KS, or, more likely, to a sexually transmitted KS cofactor that may have been more highly concentrated in these areas at this early point in the epidemic. The present study supports an independent association with use of nitrite inhalants, which could be hypothesized either to have an independent biologic effect on KS or to enhance the efficiency of transmission of the cofactor virus.