Abstract: With the rapid development of intelligent vehicles，the Safety of the Intended Functionality（SOTIF）is becoming increasingly important.As a critical component of autonomous driving systems，the automatic lane change control system poses inherent risks in decision-making and planning.This paper proposes an analysis process for the SOTIF of a vehicle lane-change decision-making system，which is based on ISO 21448 and the Systems-Theoretic Process Analysis（STPA）methodology.By integrating multiple perspectives，such as vehicle type，speed，and road conditions，the analysis process derives safety goals and improves algorithms from scenarios.The evaluation of the target vehicle’s acceleration to assess the safety of the current lane change is conducted by Gaussian process regression and fuzzy comprehensive evaluation.The optimal lane-change trajectory is determined based on the minimization of the lane change time and the desired lane endpoint.Throughout the lane-change process，the driving states of the surrounding vehicles are continuously updated，and the current safety status of the vehicle is evaluated using a proposed safety coefficient.Different lane change measures are then taken to ensure a safe lane change or a safe return in emergency situations.Finally，a verification scenario is built to compare the risks of the system before and after functional improvements under different scenarios.Our results show the risk is significantly reduced and the safety level during the lane change process is markedly improved after functional improvements.