Soil moisture was one of the most important parameters characterizing surface dryness conditions. Its retrieval by quantitative remote sensing methods had been a challenging problem. By analyzing the sensitivity of shortwave infrared MODIS band 6 and band 7 to the changes of water, the MODIS shortwave infrared spectral feature space was constructed. According to the soil moisture variation in spectral feature space, a simple and practical MODIS shortwave infrared soil moisture index (SIMI) was put forward and validated using ground-measured 0~10cm averaged soil moisture of Ningxia plain. The results show that the coefficient of determination (R2) of both them varies from 0.39 to 0.58, and SIMI has higher accuracy than temperature-vegetation index (TVX) for soil moisture retrieval. These results support the reliability of this index for regional soil moisture retrieval. However, SIMI does not eliminate the impact of mixed-pixels and still needs further improvement and perfection.