Reconstruction of Inverse Synthetic Aperture Radar(ISAR) image from its limited number of compressive echo samples is an ill-posed problem and the quality of final image significantly depends on the noise level. In this paper，a total variation based variational model is proposed for ISAR imaging from finite number of compressive echo samples based on ISAR system signal model with stepped frequency continuous wave and compressive sensing theory. An efficient Majorization-Minimization(MM) algorithm is also developed to seek the solution of the proposed model by minimizing a sequence of quadratic surrogate penalties. Results of simulated experiments with various noise levels demonstrate that the proposed method outperforms Range-Doppler(RD) algorithm and L1 norm based method when echo Signal-to-Noise Ratio(SNR) is above 10 dB.