Based on monthly COADS SSTA dataset,linear inverse modeling(LIM) is developed to predict SSTA over tropical Indian Pacific Oceans.Prediction results of independent and dependent samples show that the RMS errors of SSTA predicted in 10~12 months ahead are smaller than SSTA mean square deviation and that the relative errors in 5 months ahead are less than 50 %.The prediction accuracy is btter during El Nino and La Nina periods than during other periods,especially during La Nina periods.The prediction accuracy of LIM is better than that of auto regression model,SSA-MEM model,climate value prediction and persistence prediction.