Reports from uncontrolled studies suggest that linezolid is associated with rates of thrombocytopenia higher than those reported in clinical studies. We assessed the risk of thrombocytopenia in 686 patients with nosocomial pneumonia who received linezolid or vancomycin for > or =5 days in 2 randomized, double-blind studies and for whom follow-up platelet counts had been measured. New-onset thrombocytopenia (platelet count of <150x10(9) platelets/L) occurred in 19 (6.4%) of 295 linezolid recipients and 22 (7.7%) of 285 vancomycin recipients with baseline platelet counts of > or =150x10(9) platelets/L; severe thrombocytopenia (platelet count of <50x10(9) platelets/L) occurred in only 1 patient in each group. Platelet counts decreased to less than the baseline level in 4 (6.6%) of 61 linezolid recipients and 5 (11.1%) of 45 vancomycin recipients who had baseline counts of <150x10(9) platelets/L. No patient had a decrease to <20x10(9) platelets/L. There were no statistically significant differences between groups in these or any other platelet assessments. Clinically significant thrombocytopenia was uncommon in our analysis, and linezolid was not associated with a greater risk of thrombocytopenia in seriously ill patients than was vancomycin.