The paper tests whether the impact of prenatal care on birthweight is contaminated by selection bias, and if so, whether adverse or favorable selection dominates. A two-stage selectivity correction model with an ordered criterion function is applied to race- and ethnic-specific data from 1984...