For existing algorithms are difficult to satisfy the applied requirements, a Bayesian estimation-based algorithm for geomagnetic aided inertial navigation is proposed for aerocraft navigation. The core idea is to convert position estimation to probability estimation. A position correction strategy is also provided in this algorithm. Simulation experiments were performed in which measured data were used. Results show that the algorithm has an average navigation accuracy of 89 meters and spend average 156 seconds for first correction the INS (inertial navigation system). In addition, the performance of the algorithm is not sensitivity to initial estimation errors.