In China, CMV, PVY and TEV transmitted by aphids have seriously impaired tobacco production in most tobacco-growing areas. In this paper, several predictive models, concerning aphid-transmitted tobacco virus disease, were formulated. Based on the related data in the period of 1980-1998, with monthly average temperature and rainfall, a long-range model was developed to predict tobacco virus disease transmitted by aphids in later June before sowing. Y=8.3846+0.6504X₁+0.6585X₃+14.4583X₁₁+5.9063X₁₂ ………………….…1. (X₁-X₈ is monthly rainfall between Oct. of the preceding year and May of the current year; X9-X16 is monthly average temperature between Oct. and May.)Model "1" can be used to predict the disease occurrence of the coming year before sowing. Mid-range models were produced according to the meteorological factors between October and May, transplanting date or population density of alatae during its peak of flight.Y=-99.0763+0.2211X₁-0.0997X₂-0.9018X₆+5.8605X₉+2.3756X₁₀+9.9803X₁₁+7.3108X₁₂-0.9290X₁₃+1.7195X₁₆+1.6336X₁₇ …………………………………….……….…2.(X₁₇ is the days from Apr. 30 to transplanting date.) Using model "2", we can make a prediction to field disease index before transplanting. Y=-75.1930+0.7539X₁+0.8089X3-0.0047X₈+13.1134X₉-9.2786X₁₀+10.4861X₁₂-1.5248X16+0.091X₁₈…………………………………………………………..…3.(X₁₈ is population density of alatae at their peak of flight.)Model "3" facilitated us to predict field disease index after the peak of aphid flight.The models were examined by T-test and showed the selected factors correlated to disease index at a significant or extremely significant level and the models significantly related to disease indexes.Fitting values and errors of samples using the three models, the rate of prediction accuracy was 83.29%, 98.60%, and 99.99% respectively. When predicting the year 1999, the rate of accuracy was 96.45%, 92.75%, and 93.39% respectively. This demonstrated the accuracy of the models.We can use the models to make a long-range prediction (year prediction) or a mid-range one (about 30 days).