The grid algorithm is an attractive one in star identification algorithms. However, this algorithm needs at least 6 stars in sensor image, which limits its application to the stellar sensor whose field of view (FOV) is small or stellar magnitude limit is low. To break this limitation, a new algorithm named extended grid algorithm in star identification field is proposed. The ideas of the new algorithm is the combination of extended image algorithm and the grid algorithm, which can extend the FOV effectively, and then more stars will fall into the extended FOV. Additionally, the new algorithm inherits the advantages of grid algorithm, and its performance such as the stability and identification rate will be better than the extended image algorithm. To verify the performance of this new algorithm, a simulation according to the real situation is carried out. The results indicate that the star identification success ratio of the proposed method is larger than 97.4％, it is better than the traditional extended image algorithm, and the neighbor star identification success ratio of the proposed method can reach 86.7％ which is also better than that of the traditional grid algorithm. Therefore, the proposed method is more applicable for the small FOV or low limiting magnitude stellar sensor.