Abstract: The truncated particle filter was proposed based on the analysis of residual particle filter and regularized particle filter. The main idea of the truncated particle filter was to draw the new particles from the resampling area of the particles with large weights, rather than point-wise determine the repetition number of each particle. The effective resampling areas were established by these particles whose weights were larger than the truncated value. The uniform kernel was used to draw new particles from these areas. This method combined the information contained in the prior transformation function and the likelihood function, meanwhile increased the particle diversity. The simulation results showed that the truncated particle filter reduces the computational complexity, meanwhile maintains the same estimation accuracy as the common resampling algorithms. Furthermore, this new algorithm greatly shortened the estimated time and improves the stability of the estimates.