To evaluate factors that may affect the timely diagnosis of children with human immunodeficiency virus (HIV) infection, we compared data derived from two population-based pediatric HIV studies. Data from anonymous newborn HIV serosurveys were used to estimate the number of children born to HIV-seropositive mothers. A statewide active surveillance project determined the number of HIV-exposed children who had been clinically recognized. Of 88,732 Massachusetts newborn specimens tested anonymously for HIV antibodies during a 12-month period (November, 1987, to October, 1988), 223 were positive. As of October, 1991, 78 of these children (35%) had been identified by a statewide network of infectious disease physicians. HIV-exposed children born in inner city hospitals were more likely to have come to medical attention than those born in suburban hospitals (47% vs. 17%). Among the 29 children with confirmed HIV infection (13% of 223), the initial evaluation for HIV occurred at an earlier age among children born in inner city hospitals than among children born in other areas. HIV testing practices that rely heavily on risk assessment may result in delayed diagnosis of HIV infection in children whose mothers are not perceived to be at risk.