Persons attending outpatient clinics at Bellevue Hospital Center in Manhattan, New York City were invited to be tested for antibodies to human immunodeficiency virus (HIV). In pretest counseling, males were asked if they had injected nonprescription drugs or engaged in sex with other men since January 1, 1977; if so, they were asked not to participate and were referred elsewhere for testing. Face-to-face interviews and HIV testing were completed for 1,119 subjects with no prior indication of HIV seropositivity. Willingness to participate in the study was significantly greater among women than men and among younger than older persons. After exclusion of two subjects with indeterminate HIV serology, seroprevalence was 6.3% (70/1,117) overall, 7.1% (26/368) among men, and 5.9% (44/749) among women. HIV seropositivity among female i.v. drug users was 37% (27/74). Among heterosexuals without other HIV risk factors, estimated seroprevalences were 2.7% (25/924) overall, 3.3% (10/305) among males, and 2.4% (15/619) among females. Among heterosexual men, the data suggested associations of HIV seropositivity with sex with prostitutes and sex with numerous partners. Multiple logistic regression analysis indicated that significant predictors of HIV infection among women were a history of sexual contact with a male intravenous drug user, recent last use of intravenous drugs, and long duration of residence in New York City. Sexual intercourse with persons from AIDS risk groups was reported by 11% of men and 18% of women. The modal method of birth control reported by both men and women was "no method".