An incomplete rainfall data series could affect the reliability of related hydrological modelling. As Malaysia experiences a tropical climate with sensational variations, the availability of complete rainfall series is important for climate change assessments and water resources management. In this study, the estimation of missing rainfall data was carried out using the approach that covers an artificial neural network (ANN) and other conventional methods. These conventional methods were the inverse distance weighting method (IDW), the linear regression (LR) method, the normal ratio (NR) method and the ordinary kriging (OK) method. The performances of the estimation methods were evaluated by the goodness of fit tests, namely the mean absolute error (MAE), mean bias error (MBE), mean square error (MSE), scaled mean square error (SMSE) and the linear correlation coefficient (LCC). From the results, ANN was found to be the overall best estimation method. ANN resulted in lower values for MAE, MSE and SMSE, less biasedness for the MBE and the highest correlation for LCC. From the conventional method list, the OK was selected as the better option. Overall, ANN was more efficient approach to the estimation of missing rainfall data for the Kelantan River Basin in tropical Malaysia.