Abstract: In view of the low precision of the accelerometer, a model based on the multi-pose of the mobile phone was proposed to solve the error of the MEMS accelerometer system. Error values were obtained through solutions of n-dimensional nonlinear equations under different attitudes.The accuracy of acceleration data was improved significantly through applying the correction model. The correction result shows that the total acceleration number range of accelerometer reduces to 9.78~9.86 m/s² from 9.56~10.10 m/s² in static status, which is more close to the real gravity acceleration. The result provides a more accurate identification method for calculating displacement by step length in indoor positioning.