Objective. To compare the length of stay (LOS) and re-admissions for common pediatric medical admissions according to physician practice setting and hospital type. Design. Retrospective analysis of LOS and re-admissions for academic versus non-academic physicians and for a children's hospital versus general hospitals. Setting/sample. Alabama State Medicaid claims for pediatric patients hospitalized between 1 July 1989 and 30 June 1993 were examined for three diagnoses defined by ICD-9-CM codes: gastroenteritis, 'rule out sepsis', and urinary tract infection. Methods. Ordinary least squares regression analyses were performed for each diagnosis. Independent variables included in each model were: hospital type (children's or general), the physician's primary practice setting (academic or non-academic), the index diagnosis as principal or secondary, the presence of complicating medical conditions, Medicaid eligibility, a migration factor, patient's sex, and patient's age. Re-admission rates were compared using the Chi-square statistic. Results. After controlling for covariates, LOS for patients with the diagnoses of 'rule out sepsis' and urinary tract infection did not differ significantly for academic versus non-academic physicians. Patients with the diagnosis of gastroenteritis, cared for by academic providers, had a shorter LOS than those cared for by non-academic providers. There were no consistent differences in LOS for children cared for in a children's hospital versus general hospitals. Re-admission rates did not differ by physician practice setting or hospital type. Conclusions/implications for practice. For three common pediatric medical diagnoses, children cared for by academic providers or in a children's hospital did not have consistently longer LOSs than did other patients. For routine hospitalized pediatric patients, academic providers and children's hospitals may be as efficient as other providers.