This study was designed to examine the scale quality of our newly developed short behavior scale and to identify its role in predicting the functional outcome of stroke patients. The short behavior scale consists of six items that assess cognitive function related to the daily behavioral status of a patient in activities of daily living and exercise. It can be scored quickly through observation of a patient's behavior. We assessed 190 stroke inpatients, who had a mean age of 61 years. The mean days and length of stay from onset were 47.3 and 138.2, respectively. We examined internal consistency with the Cronbach's alpha coefficient. Interrater reliability was tested by having two examiners evaluate 30 patients independently. We studied how the short behavior scale correlated with the admission Functional Independence Measure, Mini-Mental State Examination scores, and speech and visuospatial functions. We also studied how the short behavior scale contributed to the prediction of discharge Functional Independence Measure raw scores with stepwise multiple regression analysis. In another group of 116 patients, we cross-validated our predictive equation. The Cronbach's alpha coefficient was 0.88. The intraclass correlation coefficient was 0.84 for total score. The short behavior scale correlated significantly with cognitive Functional Independence Measure scores and Mini-Mental State Examination scores. We could explain 70.8% of the variance of discharge Functional Independence Measure scores from the Short Behavior Scale, days from onset to admission, age, speech scores, and admission Functional Independence Measure scores. Stability of the predictive equation was shown in cross-validation to a second sample of 116 patients. In conclusion, our newly developed short behavior scale proved reasonable and would be useful to enhance the precision of outcome prediction in stroke rehabilitation.