It is important to evaluate walking stability to improve people's health. There are many cases of unstable walking even if people think they are in good health. A small, light acceleration sensor was attached to subjects' center of gravity to measure the acceleration displacement while the subjects were walking. By using the seasonal adjustment model it was possible to predict the periodic fluctuations observed, decomposing the original data into factors representing stability and instability. We suggest the walking stability of each subject using the ratio of the variance of instability to the variance of stability. This study provides useful information for understanding walking systems in preventive medicine and rehabilitative medicine.