As for the low centroid precision of blurry star spots acquired by the star sensor in vibration environment, the star spot centroiding model with inertial measurement information was proposed and the centroiding algorithm was formed. The weighted center of the star spot was regarded as the centroid at the initial time of the exposure by the algorithm, then the centroid track could be computed with the vibration information by the strap-down inertial sensor, and the star imaging area and its weighted center could be estimated with the star spot power distribution model. Finally, the exact centroid value at the finish time of the exposure was got by using iterative operation to make the two weighted centers coincide with each other. The centroiding precision of the algorithm and the effect of relevant factors on it are analyzed by using simulation data, and numerical results indicate that the method can reach about 0.01-pixel centroid precision even in the vibration environment with a good applicability performance.