For years, the flash floods, mudflows and landslides brought by typhoons always cause severe loss of property and human life. For this reason, it is crucial to develop a more accurate and prompt typhoon rainfall prediction technique and thus can provide necessary rainfall potential information to the relevant disaster mitigation agencies. Kidder et al. (2005) developed the Tropical Rainfall Potential (TRaP) technique, which applied satellite-borne passive microwave radiometers, to retrieve a tropical cyclone's rainfall amount and predict its 24-h accumulated rainfall distribution. However, the effects of a tropical cyclone's rainband rotation and intensity variation were not considered in their method. To obtain a better approximation to the actual rainfall system, this study will improve the TRaP technique by considering those effects. In the typhoon intensity variation part, the method proposed by DeMaria (2006) was applied to predict the 6-h intensity change with GOES-9 and MTSAT satellites, and the result was further extended to predict the 24-h intensity change and accumulated rainfall. After comparing the predicted rainfall with the rain gauge data gathered from Taiwan's offshore small islands, it shows that the accuracy of the predicted accumulated rainfall could be improved significantly while considering the effects of rainband rotation and intensity variation.