Abstract: The occurrence pattern of Spodoptera litura was studied according to the continuous year data investigated in tobacco fields and weather data of Chenzhou and Ningyuan in Hunan. The results showed that, based on effective accumulated temperature, it was predicted that Spodoptera litura occurred 4-4.5 generations each year, which was in agreement with the observed value in field in Hunan Province. The initial resources of Spodoptera litura in early stage of the year were migration pest, there were four occurrence stages in a year according to the properties of pests resources, the number of trapped moth from April to June was significantly correlated with the number in March, the forecasting model was (r = 0.815*, n = 8). The number of trapped moth from April to June was significantly correlated with the daily average temperature of January (reliability degree was 90%), the forecasting model was (r = 0.595, n = 9). The number of trapped moth of Agrotis ypsilon at its major migration stage (February-April), which synchronously migrated with Spodoptera litura, was significantly correlated with the number of Spodoptera litura in its major migration stage (April-June), and it provided data for the prediction of occurrence trend of Spodoptera.litura. The damage percent was significantly correlated with the number of Spodoptera litura per 100 plants.