Abstract: Presents a Least Squares frequency measurement algorithm based on the parallel Kalman filter. The convergence values of the Kalman filter is in accordance to the input model of the Least squares algotithm, so Least squares algorithm can be used to optimization the output of the Kalman filter. The frequency estimation accuracy can be increased significantly by using this algorithm to only using Kalman filter. It is proved that the frequency estimation error can be reduced to about 1／10 when using the Least Squares frequency measurement algorithm based on the parallel Kalman filter instead of only using the Kalman filter.