Increased physical activity is required in patients with stroke that are hospitalized in the rehabilitation unit. This study investigated the association between the daily number of steps and walking independence in order to determine the cutoff value of daily number of steps that can predict walking independence in hospitalized patients with sub-acute stroke. This cross-sectional observational study included 85 stroke patients admitted to the rehabilitation unit. The average daily number of steps was measured using Fitbit One for 4 days starting at 30 days after stroke onset. 6-min walk test, and Fugl-Meyer assessment of the lower extremities were measured The category of walking independence was classified using the Functional Ambulation Category (FAC). The subjects were divided into two groups according to the FAC score: a walking independence group (FAC ≥ 4) and a walking non-independence group (FAC ≤ 3). Logistic regression analysis was conducted to investigate the association of daily number of steps with walking independence and a receiver operating characteristic curve was used to identify the cutoff value of daily number of steps for predicting walking independence. The daily number of steps (per 1000 steps) was independently associated with walking independence (odds ratio (OR); 2.53, 95% confidence interval (CI); 1.40–5.73, p = 0.009). The cutoff value of daily number of steps for predicting independent walking was 4286 steps (area under the curve = 0.914, sensitivity of 0.731, and specificity of 0.949). The daily number of steps was associated with independent walking in hospitalized patients with sub-acute stroke. The daily number of steps may be a useful target in rehabilitation for patients with sub-acute stroke.