Abstract: This paper describes how the FFT (Fast Fourier Transform)technique is applied to the spectrum analysis of the vibration signals recorded in starting process of a turboprop engine and what features of its spectrum have been found out. Based on analysing the vibration data recorded from 174 starting processes, the basic rules of vibration in starting of the engine are found and essential difference between normal and troubled engines is clarified. Vibration divergence in hot and cold startings and the influences of thermal deformation of the engine rotors on the vibration are also investigated.This work has accumulated data and experiences for fault diagnosis and monitoring in starting of the engine.