The purpose of this study was to test, refine, and extend a statistical model that adjusts neonatal intensive care costs for a very low birth weight infant's day of life and birth weight category. Subjects were 62 infants with birth weights below 1,501 g who were born and cared for in a university hospital until discharged home alive. Subjects were stratified into 250-g birth weight categories. Clinical and actual daily room and ancillary-resource costs for each day of care of each infant were tabulated. Data were analyzed by using a nonlinear regression procedure specifying two separate for modeling. The modeling was performed with data sets that both included and excluded room costs. The former set of data were used for generating a model applicable for comparing interhospital performances and the latter for comparing interphysician performances. The results confirm the existence of a strong statistical relationship between an infant's day of life and both total hospital costs and the isolated costs for ancillary-resource alone (P < 0.0001). A refined series of statistical models have been generated that are applicable to the assessment of either interhospital or interphysician costs associated with providing inpatient care to very low birth weight infants.