We propose a new method for evaluating spasticity of ankle joint muscles by using microvibration (MV) measurement. In order to measure the MV, a piezo-electric pick-up sensor was used and set on the M. gastrocnemius. The MV signal was detected in both sitting and standing positions, and was recorded. Thereafter, the MV signal was analyzed by digital computer. The subjects consisted of two groups: (1) 11 male healthy individuals and (2) 10 male and 1 female hemiplegic individuals with spasticity of ankle joint muscles. The MV signal was subjected to Fast Fourier Transform (FFT), with sampling frequency of 500 [Hz] and 1024 data points. For estimating the difference of MV characteristics between healthy and spastic subjects, the following analyzing method was employed: First, FFT amplitude spectra of the MV signals in sitting and standing positions were obtained and the latter amplitude spectrum was subtracted by the former amplitude spectrum. Next, low frequency band component (L. F. B.) and high frequency band component (H. F. B.) were obtained from the subtracted spectrum. Ratio of both components to the total spectral power were calculated as the percent expression (L. F. B. C. [%] and H. F. B. C. [%]). Furthermore, L. F. B. C. [%] and H. F. B. C. [%] were plotted on x-y coordinates, and Euclid distance between 4 healthy and 6 spastic subjects' data was calculated. We obtained maximum Euclid distance when L. F. B. having the center frequency at 5 [Hz] with the band width of 10 [Hz] and H. F. B. having the center frequency at 35 [Hz] with the band width of 30 [Hz] respectively. With this discrimination method, MV characteristics of 11 healthy and 11 spastic subjects were plotted on x-y coordinates. The result showed a significant difference of MV characteristics between normal and spastic subject groups. Increase in H. F. B. and decrease in L. F. B. represented decrease of spasticity. This was also verified and correlated with a phyisio-therapist's evaluation. Furthermore, continuous change of spasticity in standing position on an oblique board was found by using this method.