Abstract: he teleconnection between population density (egg counts) of cotton bollworm (CBW), Helicoverpa armigera （Hübner）, and the anomaly of monthly mean sea surface temperature (SST) in the North Pacific was analyzed to find the significant (P<0.05) correlative key factors with large and sustainable spatiotemporal extension to build longterm forecast models for Helicoverpa outbreaks. Based on the CBW data over 26 years (1974-1999) in Yuncheng, Shangdong Province, over 22 years (1978-1999) in Dezhou, Shandong Province, and over 20 years(1980-1999) in Fengxian, Jiangsu Province, optimum forecast models were established for longterm prediction of the 3rd generation eggs in Yuncheng and Dezhou, and the 2nd generation eggs in Fengxian. The results indicated:（1）There was significant (P<0.05) or highly significant (P<0.01) teleconnection between the fluctuation of CBW populations and SST in the North Pacific. The location and range of the significant correlated areas varied with time, but there were a few such significantly correlated areas, relatively stable in space and sustainable in time. （2）There was a vast region in the North Pacific of significantly correlated (P<0.001 in center) with the CBW eggs counts of the 3rd generation in Yuncheng, the 2nd generation in Fengxian and the SST in January 2 years ago, which ranged from 35°N to 55°N, 135°E to 135°W and sustained for 4 months. But the area for the 3rd generation eggs in Dezhou was located at the low latitude Northern Pacific (1°-17°N，165°E-120°W) during July to September 2 years ago.（3）The evaluation of models with the drive variable of the average of the mean SST departures on each grid within the significant correlative areas from January to November of the year before last showed that the predictive accordant rates for the egg counts of the 3rd generation CBW in Yuncheng and Dezhou, and the 2nd generation in Fengxian were correct in 5 of the 6 verified years (1994-1999) for Yuncheng, correct in all 5 years (1995-1999) for Dezhou and in 3 of 5 years (1995-1999) for Fengxian respectively. The results suggested the possibility to make forecast of CBW outbreaks about 20-27 months earlier with these models.

Key words: Helicoverpa armigera, sea surface temperature, correlation analysis, l ongterm outbreak forecast.