In this study, we used an analytic hierarchy process (AHP) to establish a prediction index system model for medium-strong earthquakes in middle-eastern Tianshan, Xinjiang, based on better predictors of earthquakes in that region. On this basis, we built a judgment matrix with a 1~9 ratio scale and then calculated the weight of each index. The results indicate that among the better predictors reflecting earthquakes, those with relatively large weight values reflect deformation anomalies and some seismometry parameters can be used as short-term prediction indicators. The AHP model also provides a comprehensive prediction index Y for medium-strong earthquakes in middle-eastern Tianshan, whereby the Y value can better reflect the strength or weakness of the earthquake precursor and abnormality reliability. When the value of Y is no less than 1.15, medium-strong earthquakes are likely to occur in middle-eastern Tianshan, Xinjiang in the short term.