To establish the seroprevalence of the human immunodeficiency virus (HIV) in our county hospital obstetric patients, and to assess the predictive value of screening questionnaires for high-risk behaviors attributing to HIV infection, we conducted written surveys and blinded HIV testing over a 4-month period ending February 1993. Blinded HIV antibody testing was performed on 1348 patients upon admission to labor and delivery. The coded blood samples were matched to similarly coded surveys of patient demographics and behaviors implicated in HIV transmission. The overall HIV prevalence in our regional population remains virtually unchanged at a rate of 0.22% (2.2 per 1000 patients). Questionnaires regarding health history were not predictive for HIV infection in our patients. Universal screening questionnaires cannot predict HIV infection. Similarly, routine mandatory testing for HIV performed on obstetric patients on the day of delivery are not cost-effective and do not provide clinically meaningful information needed to alter management protocols.