After contradictory findings from a number of previous studies, behavioral risk factors for Kaposi's sarcoma were examined in a case-control study of 202 people diagnosed with acquired immunodeficiency syndrome (AIDS) in 1991-1993 in Sydney, Australia. Cases comprised 67 men who developed Kaposi's sarcoma at or after a diagnosis of acquired immunodeficiency syndrome, and controls were 135 people who did not have Kaposi's sarcoma at the time of diagnosis of acquired immunodeficiency syndrome or during follow-up until 1995. Men who developed Kaposi's sarcoma were more likely to report having a history of sexually transmissible diseases and having engaged more frequently than controls in a number of sexual practices with casual partners in the period before they became aware of their human immunodeficiency virus (HIV) infection. However, the only sexual practice reported significantly more often by cases at the 0.05 significance level was insertive oroanal contact with casual partners (odds ratio = 2.6, 95 percent confidence interval 1.3-5.3). This association was not present for insertive oroanal contact with regular partners or for insertive oroanal contact after subjects became aware of their HIV infection. The relation was present both in men who had Kaposi's sarcoma at the time of interview and in those who developed it later. The relation was not affected by adjustment for time of HIV infection and diagnosis or for other sexual practices. These results can be interpreted as supporting the hypothesis that Kaposi's sarcoma in people with HIV is caused by an infectious agent transmitted by oral contact with feces.