Vancomycin trough levels are recommended to predict vancomycin efficacy, and inaccurate levels may lead to inappropriate clinical actions. However, the frequency of timing errors and associated clinical impact is unknown. We retrospectively analyzed vancomycin levels (n = 2,597) measured during 13 months at a large academic medical center. Of the specimens, 41.3% were drawn too early. These samples yielded significantly higher average ± SD vancomycin concentrations than correctly timed samples (22.1 ± 11.7 mg/L vs 15.5 mg/L ± 8.6 mg/L; P < .001), and, consequently, clinicians were more likely to decrease, discontinue, or hold a patient's vancomycin dose (25.6% vs 21.4%; P < .02) or repeat the vancomycin level (29.2% vs 20.0%; P < .001). A substantial proportion of specimens collected to assess vancomycin efficacy were drawn too early, leading to overestimation of patients' true trough level and possible underdosing of vancomycin or a high rate of repeat tests for vancomycin.