Achieving the strategical goals in forestry of Republic of Serbia will not be easy in the light of climate change. Gypsy moth (Lymantria dispar L.) is the most economicaly significant and aboundat pest in deciduous forests in Serbia. It is also very important pest in fruit orchards. His outbreak often has the character of a natural disaster that requires a significant commitment of manpower and financial resources in order to supress it. We developed two models for predicting the occurrence of gradation (outbreak) and latency of gypsy moth population on the basis of monthly and quarterly values of climatic data for the period 1888-2010. The models were based on logistic regression. In the MODEL I, we have used the mean monthly temperatures from October of the year preceding event, temprature in January and March, and the rainfall in May, while in MODEL II, taken were mean temperatures of the first quarter and sum of the precipitation of the second quarter. Overall classification accuracy of the models were above 70%, while the prediction of outbreak based on MODEL I was 86%. The results of this study (models that can be applied in real time) can contribute to better decision-making in relation to forest management and protection of forests from gypsy moth in Republic of Serbia and wider.