The use of neonatal intensive care (NIC) continued to rise rapidly in the 1990s despite the concerns of observers about its cost effectiveness and its successes being mostly in facilities with high volume and capabilities. The objective of this study is to test the effects of insurance type, competition among hospitals, and market pressure from managed care plans on the supply and cost of NIC. The analysis uses logistic and linear models with techniques to avoid bias from (a) market area definitions based on actual patient flows and (b) self-selection of hospitals by patients with unmeasured risk of needing NIC. The data source contains all births in short-term hospitals in New Jersey during 1990 and 1994. Both the number of days and charges for NIC are reported. Key findings are that the decision of a hospital to offer NIC was associated with teaching status, the proportion of infants in the market area with documented high risk, and the market concentration of major competitors. The market share of managed care plans and the concentration of enrollment were not associated with either NIC being offered or with the standardized charges. Whether a particular patient was given to a NIC depended on patient risk factors and whether a NIC unit was present, but not on payer group. The results are consistent with the hypothesis that young insured parents (with the advice of their obstetricians) prefer hospitals with NIC and also are relatively profitable enrollees for health plans. In conclusion: using the results here and in other research, public and private policy makers may consider several ways to strengthen the incentives for health plans to contract for cost-effective birth-related services. The results also raise questions for a number of regulatory and payment policies and call for better public data on costs and outcomes for NIC.