Improving the positioning accuracy of star spots is still one of crucial technologies in high precision star trackers. The minimum meansquared error of the conventional center of gravity (COG) and Gaussian curve fitting (GCF) algorithm for positioning is about 0.1 pixel, hardly to meet the future demand for highprecision positioning. In the field of adaptive optics, an iteratively weighted center of gravity (IWCOG) is proposed for wave- front localization of Shack- Hartmann sensor with superior performance to the first two algorithms in locating accuracy. However, the positioning mechanism of IWCOG and the performance on the multi-star positioning accuracy are currently unknown. On the one hand, this paper introduces the positioning mechanism of the algorithm from a theoretical perspective of Meanshift and proves the convergence of the IWCOG algorithm. On the other hand, 10000 randomly star spots are generated with Monte Carlo method in every single simulation to study the influence of IWCOG algorithm on multi- star positioning accuracy in star tracker by studying relationship between the star point parameters, signal to noise ratio and positioning errors. The result shows that there are 44% samples with positioning error lower than 0.02 pixel and 72% with error lower than 0.04 pixel using IWCOG algorithm under signal to noise ratio R equal to 40, much better than the COG with only 0.8% and 1.9% , basically meeting the positioning requirements of super- high accuracy star tracker.