Clinical and billing data were collected on all admissions to six California newborn intensive care units during a 6-month period. Charges were adjusted to costs using Medicaid cost to charge ratios and for inflation, and patients were classified by the diagnosis-related group (DRG) system. Costs were from 97% to 708% more than the proposed DRG reimbursement levels. Regression analysis showed that DRGs explained 22% of the variation in costs. An alternative model using binary variables to control for birth weight, assisted ventilation, surgery, survival, multiple births, and mode of discharge explained 42% of the variation in costs. In contrast to other proposed DRG alternatives, this simple model does not require special training or subjective decision-making.