Abstract: On the besis of steppe grass yield during 1966--1987 in Ziniguan stud sheep farm and meteorologic data of shawan station, the grass yield variation and the effect of meteorologic conditions on grass yield were studied by statistical methods of correlation analysis, integral regression and stepwise regression analysis. The results showed that the grass yield does not increase or decrease with time, the percipitation is a key factor. April—June and September—October were key periods. Finally, a grass yield predicting model was established.