OBJECTIVES: To investigate variation among neonatal intensive care units (NICUs) in prevalence and management of thrombocytopenia in infants and(SNAP). Platelet counts in the first 12 hours after birth and on day 3 of life were abstracted from the infants' medical records. Thrombocytopenia was determined from the lowest platelet count in each of these time periods. RESULTS: There was variability in rates of thrombocytopenia among NICUs, even after controlling for risk factors (e.g., SNAP, small for gestational (SGA) age and maternal hypertension). One site had a high prevalence of thrombocytopenia, but the lowest percentage of infants with thrombocytopenia who received platelet transfusions. After controlling for SNAP, GA, SGA, Apgar score and incidence of thrombocytopenia, the odds of receiving platelets at this site, relative to the site with the highest transfusion rate, was 0.10 (95% CI 0.02 to 0.43). CONCLUSIONS: This multicenter study finds a 10-fold variation among NICU in the administration of platelets to their thrombocytopenic infants that cannot be explained by presence of thrombocytopenia or illness severity.