The invention provides a vertical speed estimation method of an all-wheel drive vehicle. The method comprises the following steps of firstly, collecting data based on an acceleration sensor and a gyroscope, using a Kalman filtering algorithm to realize multiple sensor data fusion, and acquiring estimation of a current driving road grade of a vehicle; then, based on a kinematic vehicle speed estimation method, estimating a vertical driving speed of the vehicle according to a grade estimation result and acceleration sensor data; and finally, eliminating an integral error caused by a noise in acceleration data, and correcting a vehicle speed estimation result according to each wheel speed feedback. In the invention, through combining with a road grade identification result, the vertical speedestimation is obtained by using a kinematic estimation method, and the vehicle speed estimation result is corrected according to each wheel speed feedback so that the vehicle speed estimation of thefull-drive vehicle can be realized. The method has versatility, high real-time performance and high calculating precision. Accurate vehicle state information can be provided for a vehicle electronic control system.