Maternal exposure to illicit substances during pregnancy may have detrimental effects on newborns that increase the costs associated with caring for these infants immediately after birth. Yet accurate estimates of the neonatal costs of prenatal drug exposure have been difficult to ascertain. To minimize selection bias in constructing these estimates, one must adjust for the influence of many demographic and behavior variables. This necessitates the use of large data sets. To estimate the neonatal costs of prenatal drug exposure in New York City, we analyzed discharge abstracts linked to birth certificates for all infants delivered at the city's 11 municipal hospitals between 1990 and 1992 (N=72,899). Exposures recorded on discharge abstracts or birth certificates were coded for cocaine alone, opiates alone or cocaine plus opiates. The cost of each discharge was calculated using the DRG-based New York Prospective Hospital Reimbursement Methodology. In log-linear multivariate analyses, we controlled for a spectrum of demographic, socio-economic, obstetric and prenatal care variables. We found that, compared to the geometric mean of costs associated with unexposed newborn infants($1,827), costs associated with prenatal exposure to cocaine alone were 85% higher ($3,377, p<.01), costs associated with exposure to opiates alone were 286% higher ($7,049, p <.01), and costs associated with exposure to cocaine and opiates were 320% higher ($7,666, p<.01). We conclude that, even after controlling for the selection bias associated with multiple demographic, socio-economic, obstetric and prenatal care variables, there are significant incremental costs of neonatal hospitalization associated with prenatal exposure to illicit substances.