This study assesses the inter-annual variability of summer monsoon rainfall of Northern Hilly Area of Pakistan, which includes Balakot, Chitral, Murree, Gilgit, Skardu, Chilas, Muzaffarabad and Dir. The region receives a heavy rainfaU, the total annual rainfall being 1000 mm or more. We attempt to model the rainfall process of the data from 1971-2000. Initially we use twelve predictors out of which only four, viz., sea surface temperature (SST-lOa), temperature of Hyderabad city of Pakistan (HT-5)and pressure of Kakul (KP-10)and Lahore (LP3), are selected with the help of stepwise multiple linear regression. The mean square error of the regression model is found to be 13.8%. Since the entire modeling procedure is based on selected predictors, we suggest that the selected predictors, which are selected here according to their correlation, should be refined by using more sophisticated technique such as principal component analysis, or nonlinear correlations.