OBJECTIVE : The association between prestroke sarcopenia and stroke severity has not been evaluated previously. The purpose of this study was to determine whether prestroke sarcopenia is associated with stroke severity in elderly patients with acute stroke. METHODS : We assessed prestroke sarcopenia of elderly patients with acute stroke by using a questionnaire for sarcopenia (SARC-F). Patients were divided into groups according to their SARC-F score: SARC-F score less than 4 (nonsarcopenia) and SARC-F score ≥4 (prestroke sarcopenia). Stroke severity was assessed according to the National Institute of Health Stroke Scale. Logistic regression was used to derive crude and adjusted odds ratio for the presence of prestroke sarcopenia and stroke severity. RESULTS : Among the 183 patients enrolled (age, median [interquartile range]: 75  years; 103 men), the prevalence of prestroke sarcopenia was 15% (n = 27). Crude odds ratio for the presence of prestroke sarcopenia and moderate-to-severe stroke (National Institute of Health Stroke Scale score > 5) was 4.00 (95% confidence interval, 1.68-9.53; P = .002). After adjusting for confounding variables (age, sex, and stroke risk factors), the presence of prestroke sarcopenia remained an independent predictor of severe stroke, with an odds ratio of 3.54 (95% confidence interval, 1.32-9.49; P= .01). CONCLUSIONS : Prestroke sarcopenia can predict moderate to severe stroke in elderly patients with acute stroke.