Abstract: In order to study the accuracy and applicability of Markov grey residual GM（1,1） model in predicting compressive strength of cement pisha sandstone, after a series of processings of compressive strength data, the grey GM（1,1） model and grey residual GM（1,1） model were established. Then the Markov grey residual GM（1,1） model was constructed based on the Markov process, and the compressive strength of cement-cured pasha sandstone was estimated by this model. The results show that the test accuracy of the grey residual GM（1, 1） model is greatly improved and all the test indexes basically reach level 1, which is obviously better than that of the grey GM（1, 1） model. The Markov process is convenient to determine the positive and negative signs of residual correction, and then the compressive strength of cement solidified Pisha sandstone with different cement contents at 90 days was predicted by the Markov grey residual GM（1,1） model. The relative error is reduced from 1.77%-4.01% to 0.60%-2.36%, and the average relative error is reduced from 2.63% to 1.25%. The prediction accuracy of the model is improved obviously. This study can provide a simple and reliable new method for predicting the compressive strength of cement-cured Pisha sandstone and other cement-based engineering materials.