The sudden transition to rapid weakening immediately after a tropical cyclone (TC) reaches its intensity peak (hereafter TP-RW) is a challenge to operational TC intensity forecasts, but the characteristics and the controlling factors of the TP-RW have not been investigated so far. In this study, the main characteristics of the TP-RW events over the western North Pacific and the corresponding key controlling factors and their relative importance are analyzed and quantified. Results show that about 2.5 TP-RW events occurred each year, which corresponds to about 10% of total TCs over the western North Pacific during 1982–2018. The TP-RW events occurred in all months but peaked in fall and were the most likely to occur for TCs with intensities between 90–145 kt. The sea surface temperature, the TC maximum potential intensity and translational speed, the environmental vertical wind shear and relative humidity are key to the TP-RW. Four distinct clusters of the TP-RW events are identified based on their tracks using the K-means clustering algorithm. The relative contribution of each of the identified key factors to each cluster is quantified using the growth rate from the simplified dynamical system for TC intensity prediction developed by DeMaria. The reversal of the growth rate from positive to negative values is shown to be a good indicator of the TP-RW processes. Different clusters are predominantly controlled by different key factors. The findings from this study can help better understand TC intensity changes and improve TC intensity forecasts for such kind of RW. This article is protected by copyright. All rights reserved.