Prediction of Indian monsoon rainfall variability has long been a goal of seasonal forecasting. Forecasts of seasonal rainfall anomalies are feasible to the extent that these anomalies are related to predictable shifts in boundary conditions such as land surface conditions and sea surface temperature. Here we focus on the response of the climate system to sea surface temperature (SST) forcing and ask how knowledge of SST can be used to infer information about Indian monsoon rainfall variability. Computing the correlation of observed precipitation with that simulated by ensembles of general circulation models (GCMs) forced with observed SST gives an estimate of the influence of SST on expected precipitation. In the case of Indian monsoon rainfall this influence appears weak in many GCMs. However, GCM deficiencies as well as lack of predictability can cause poor simulation skill. Some model deficiencies can be corrected using statistical regression methods. Information theory provides a guiding framework for computing corrections that account for all the information contained in the ensemble forecasts and observations. Here we use statistical corrected GCM simulations to estimate seasonal and spatial characteristics of Indian monsoon rainfall predictability. Implications for seasonal forecasting and retrospective forecasts of the failed Indian Monsoon of 2002 are discussed.