As a core component of power conversion systems, insulated gate bipolar transistor (IGBT) modules continually suffer from severe thermal damage caused by temperature swings and shear stress, resulting in fatigue failure. Bond wires falling off is one of the failure modes of IGBT modules. Given that the number of fallen-off bond wires is a significant parameter to evaluate the health status of the IGBT modules, this paper proposes an online identification model to recognize the number of fallen-off bond wires during normal operation. Firstly, a database containing datum Vce,on-Tj-IC (collector-emitter on-state voltage Vce,on, chip junction temperature Tj, collector current IC) planes with different fallen-off bond wires is built based on an offline aging test. Secondly, a Foster network model and a special circuit are designed to measure the junction temperature Tj and the collector-emitter on-state voltage Vce,on, respectively. Thirdly, the feature points of the IGBT module represented by Vce,on, Tj, and IC are given to the database to recognize the number of fallen-off bond wires according to the position of the feature points in the datum plane. The experimental results show that the proposed method can determine the fallen-off bond wires under the operation condition.