Epidemiologic investigation of the AIDS epidemic among heterosexuals has consisted chiefly of studies of partners of individuals infected with the human immunodeficiency virus (HIV) and population surveillance. Heterosexual partners of infected individuals appear to be at high risk of infection, but only a small proportion of cases of AIDS have been attributed to heterosexual contact in the United States and Europe. An epidemic model for heterosexual spread of HIV infection is developed and fit to surveillance data. Fitted values are restricted to a range consistent with findings from partner studies. Because, at present, most HIV-infected heterosexuals and bisexuals have been infected through other means (intravenous drug use or homosexual contact), the model considers two interacting populations: a small population of individuals rapidly infected by high-risk activity and a large population of individuals at risk only from heterosexual contact. No precise predictions concerning the AIDS epidemic among heterosexuals are possible now, but current epidemiologic findings neither predict nor preclude a major heterosexual epidemic. Projections depend strongly on the delay between infection and infectivity. The model can also be used to demonstrate how interpretation of results of case-control studies of HIV infection depends on underlying assumptions about the dynamics of the epidemic.