It is widely believed that expanding prenatal care should improve infant health; yet research typically finds weak effects of prenatal care on infant health. We argue that there are two kinds of pregnancies, 'complicated' and 'normal' ones, and that combining these pregnancies, as past research does, may lead prenatal care to appear ineffective. Data from the NMIHS offers compelling evidence supporting our view. The standard 2SLS approach yields residuals that are obviously bimodal and prenatal care coefficients that are frequently insignificant. In contrast, estimating birth weights with a finite mixture model yields estimates that are much more robust and that clearly suggest that prenatal care has a consistent, substantial effect on ?normal? pregnancies. Our Monte Carlo experiment confirms that ignoring even a small proportion of ?complicated? pregnancies can lead the onset of prenatal care to appear unimportant.