Objective: This study was performed to examine differences in the association between straight and curved walking abilities of inpatients in the subacute phase and walking independence level. Design: Cross-sectional study. Methods: Subjects were divided into an independent group and a supervised group (n=10 each) by walking independence level within the ward decided by physical therapists. Inclusion criteria comprised the ability to ambulate independently within the ward, regardless of the use of walking aids. Straight walking abilities (walking velocity, stride length, and cadence) were evaluated using the 5-meter walk test. Curved walking abilities were evaluated using the Figure-of-8 Walk Test (F8W) and the 3-meter zigzag walk test (3ZW). Differences in associations between straight and curved walking abilities of inpatients were examined by calculating correlation coefficients between straight and curved walking abilities. Results: Age, walking velocity, stride length, F8W and 3ZW varied markedly between independent and supervised groups. In the independent group, F8W and 3ZW correlated significantly with walking velocity and cadence (p<0.05). In the supervised group, F8W correlated significantly with walking velocity and stride length (p<0.05), but 3ZW did not correlate significantly with straight walking abilities. Conclusions: The association between straight and curved walking abilities varied between inpatients who could ambulate independently and inpatients requiring supervision for ambulating. These differences may depend on skill on straight and curved walking abilities. There is a possibility that curved walking exercise is necessary for supervised group.