We conducted a prospective cohort study of 323 consecutively born very low birth weight infants (< or = 1499 gm) to determine any association between prenatal cocaine exposure and (1) intracranial ultrasonographic abnormalities and (2) other adverse perinatal outcomes. The infants were assigned to either a cocaine-exposed group (n = 86) or a cocaine-nonexposed group (n = 146) on the basis of combined detection methods for prenatal maternal cocaine abuse including maternal history, maternal and infant urine immunoassay (Emit), and meconium analysis (high-performance liquid chromatography and gas chromatography-mass spectrometry). Ninety-one infants were not assigned because of early death before complete testing (n = 80) or missed tests (n = 11). The detected incidence of cocaine exposure in the assigned population was 37% (86/232). Meconium testing with high-performance liquid chromatography and gas chromatography-mass spectrometry was the sole means of detection in 30% (26/86) of cases. The cocaine-nonexposed infants did not differ from the cocaine-exposed infants in the incidence of intraventricular hemorrhage (36% vs 35%), grades III and IV intraventricular hemorrhage (14% vs 14%), or periventricular leukomalacia (4% vs 2%). Adverse outcomes increased by cocaine exposure were abruptio placentae (8% vs 18%; p = 0.046), surgical ligation of a patent ductus arteriosus (1% vs 7%; p = 0.02), and seizures (5% vs 17%; p = 0.004). We conclude that prenatal cocaine exposure does not increase the incidence or severity of intracranial hemorrhage or periventricular leukomalacia but does increase the risk of abruptio placentae, surgical ligation of a patent ductus arteriosus and seizures in very low birth weight infants.