BACKGROUND Human immunodeficiency virus type 1 (HIV-1) infection is known to increase the incidence of Kaposi's sarcoma and non-Hodgkin's lymphoma. Parallels with other causes of immunodeficiency suggest a possible effect of HIV-1 on additional cancers. OBJECTIVE This study was designed to determine the types and rates of cancers occurring in excess in the presence of HIV-1 infection. METHODS We examined cancer incidence in a population-based open cohort with a high prevalence of HIV-1 infection. The study population was never-married men aged 25-54 years who resided in San Francisco, Calif., of whom an estimated 20,000 (24%) were HIV-1 seropositive as of late 1984. Cancer registration data covering 1,390,000 person-years of observation of these men from 1973 through 1990 were obtained from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program. Standardized incidence rates and ratios of observed to expected cases (based on rates in the pre-acquired immunodeficiency syndrome [pre-AIDS] period [i.e., 1973-1979]) were calculated for cancers classified by site and by cell type. RESULTS The incidence of Kaposi's sarcoma in never-married men plateaued in 1988-1990 at 0.5% per year. The incidence of non-Hodgkin's lymphoma increased 20-fold between 1973-1979 and 1988-1990; increases were most pronounced in tumors of higher grade histology and extranodal (especially central nervous system) primary sites. The incidence of Burkitt's and Burkitt-like tumors peaked in 1985-1987, whereas that of large cell diffuse and immunoblastic lymphomas increased throughout the study period. The incidence of Hodgkin's disease was 2.0 (95% confidence interval [CI] = 1.3-3.0) times expected in 1988-1990. The incidence of anal cancer was 9.9 (95% CI = 4.5-18.7) times expected in 1973-1979 and 10.1 (95% CI = 5.0-18.0) times expected in 1988-1990. Ratios of observed to expected cancers of most other sites were 2.0 or less; the ratio of leiomyosarcomas (at any site) was 2.5 (95% CI = 0.5-7.4). CONCLUSIONS As the HIV-1 epidemic has progressed, the increases in AIDS-related Kaposi's sarcoma, Burkitt's tumor, and other non-Hodgkin's lymphoma have followed different patterns. The effect of HIV-1 on other cancers has been nondetectable. In particular, HIV-1 is not related to the increased risk of anal cancer in homosexual men, which antedated the AIDS epidemic. CONCLUSIONS These data suggest that the etiologic mechanisms of HIV-1-related malignancy differ for specific cancers and do not globally increase cancer risk. Control of HIV-1-related cancer remains an unsolved challenge in the management of HIV-1 infection.