The authors compared cases of acquired immunodeficiency syndrome (AIDS) diagnosed in San Francisco, California, during 1983-1984 with human immuno-deficiency virus (HIV) antibody-negative neighborhood and clinic controls, looking for risk factors for clinical AIDS. They also compared antibody-positive with antibody-negative neighborhood and clinic controls for risk factors for HIV infection. Odds ratios were 52.0 for AIDs and 7.8 for seropositivity for more than 100 sexual partners versus 0-5 partners when antibody-negative neighborhood controls were compared with cases and with antibody-positive neighborhood controls, respectively. Odds ratios were only 2.9 and 3.4 when antibody-negative clinic controls were compared with cases and with antibody-positive clinic controls, respectively. Odds ratios of 4.6-7.3 for rectal receptivity with most or all partners versus none or one partner were statistically significant, independent of the number of partners. Douching before sex was independently associated with odds ratios of 2.2-2.8. There was no evidence for oral-genital, oral-anal, or other sexual transmission of AIDS. In multivariate analysis, independent odds ratios of 2.4-6.0 for prior syphilis and 10.8-27.9 for prior giardiasis were statistically significant or marginally significant in all comparisons. There was a moderate association with nitrite use. No other drugs were consistently associated with clinical AIDS or HIV seropositivity. Odds ratios associated with AIDS and seropositivity were closely comparable except for number of partners.