Inverse synthetic aperture radar is an imaging radar in which the radio image is derived by utilizing the Doppler effect produced by change of the target aspect. When the target is accelerated, the resolution in this radar is degraded due to the frequency modulation of the received signal. In order to avoid this, the phase of the received signal must be compensated. In general, however, the motion of the target cannot be predicted, and the amount of correction must be determined by a certain evaluation index derived from the received signal. Such a method, where the amount of correction is derived from the received signal and the resolution is improved, is generally called autofocus. In conventional synthetic aperture radar, there is a method in which the peak luminosity of the image is used as the evaluation index in compensating the frequency modulation of the received signal produced by vibration of the platform. In this article, it is proposed to use the entropy of the image, and the result is compared to the conventional method. It is shown by a computer simulation that the probability of compensation failure can be reduced to the same value or decreased to approximately 1/100 of the conventional method. In other words, it is shown that the entropy, rather than the peak luminosity, of the image should be used as the evaluation index for the autofocus. Β©