In this paper, the estimation of hybrid position for mixing GMSK signals is researched，and a iterative algorithm base on parameter estimation and hybrid position estimation is proposed. Considering the differences of mixed and non-mixed part in frequency and amplitude, the frequency and amplitude can be estimated by zero-crossing and extreme points estimation, and then, each sampling point is mixed or not can be judged. The hybrid origin is correspond to relative time delay, when the delay is large , the parameter estimation will be better when using the non-mixed part data. The CRLB(Cramer-Rao lower bound) for onset estimation of the mixing signals is also researched, the simulation shows that, the proposed algorithm has high precision but less data, can be use in burst mode ,and it shows good robustness to different parameters.