In this study, four General Circulation Models (GCMs) from Coupled Model Intercomparison Project Phase 5 (CMIP5) were applied to predict monthly precipitation over Kuching, Sarawak. A feed forward neural network technique was pursued using the Levenberg-Marquardt method to train and predict monthly precipitation. HadGEM2-AO and MIROC5 performed better than BCC-CSM1.1 and CSIRO-Mk3.6.0 when compared by correlation coefficient and root mean square error. Overall HadGEM2-AO performed better than all GCMs when compared for the monthly precipitation prediction. All models underestimated monthly precipitation during the December to February and overestimated monthly precipitation during March to May. Except HadGEM2-AO, all other models were unable to predict monthly precipitation during Jun to November. However, HadGEM2-AO was able to predict monthly precipitation more realistically in the historical run for all months. © 2016 Taylor & Francis Group, London.