Objective.To describe the epidemiology of vancomycin usage at a children's hospital.Methods.A cohort study of patients at Egleston Children's Hospital who were charged for the receipt of vancomycin from October, 1992, through October, 1995, was performed. Data were obtained from pharmacy charge records in the hospital's medical records information system.Results.During the study period there were 3589 patient hospitalizations in which vancomycin was used. Patients receiving vancomycin were predominantly male (56.6%) and white (62.4%), ranged in age from 0 to 31 (median, 3.8) years and had an average length of stay of 6.0 days. The total number of vancomycin doses was 105 704; the median number of vancomycin doses during each patient hospitalization was 11.0 (range, 1 to 1215). The total charge for vancomycin used was $2 009 746; the median charge for vancomycin per patient was $297.50 (range, $11 to 19 864). The majority (75.7%) of vancomycin doses were given on the hematology (27.6%), neurosurgery (17.9%), cardiothoracic surgery (13.4%), neonatology (9.7%) or general pediatrics (7.1%) services. Overall surgery service patients were significantly more likely to receive vancomycin than were medicine service patients (1267 doses/6221 admissionsvs.1954/19 446; relative risk, 2.03;P< 0.001). During the study period the number of vancomycin doses decreased significantly (P< 0.001).Conclusions.This study shows the value of evaluating antimicrobial use through a pharmacy database. Although vancomycin use decreased during the study period, large amounts of vancomycin are still being prescribed primarily on subspecialty service patients. Interventions to reduce vancomycin use should focus on these groups.