Accurate information of cropland area and spatial location is critical for studies of national food security, global environmental change, terrestrial ecosystem geophysics and the geochemical cycle. In this paper, we compared five global cropland datasets in circa 2010 of China from in terms of cropland area and spatial location, including Global Land30, FROM-GLC, Glob Cover, MODIS Collection 5, and MODIS Cropland. The results showed that the accuracies of cropland area and spatial location of Globe Land30 were higher than the other four products. The cropland areas of the five products varied in most of the provinces. Compared with the statistical data, the best goodness of fit was obtained using Globe Land30, followed by MODIS Collection 5 and FROM-GLC, with MODIS Cropland and Glob Cover having the poorer accuracies. Regarding the spatial location of cropland, Globe Land30 achieved the best accuracy, followed by FROM-GLC and MODIS Collection 5, with Glob Cover and MODIS Cropland having the poorer accuracies. In addition, the spatial agreement of the five datasets was reduced from agricultural production area to pastoral area and significantly affected by elevation and slope factors. Although the spatial resolution of MODIS Collection 5 was the lowest, accuracies of the cropland area and spatial location were better than those of Glob Cover and MODIS Cropland. Therefore, high spatial resolution remote sensing images can help to improve the accuracy of the dataset during land cover mapping, while it is also important to select a suitable classification method. Furthermore, in northwestern and southeastern China, spectral mixing pixels are universal because of the complicated landscape and fragmentized topography and result in uncertainty and poor consistency when using the five products. Therefore, these regions require additional attention in future cropland mapping studies. Accurate information of cropland area and spatial location is critical for studies of national food security, global environmental change, terrestrial ecosystem geophysics and the geochemical cycle. In this paper, we compared five global cropland datasets in circa 2010 of China from terms of cropland area and spatial locations, including Global Land 30, FROM-GLC, Glob Cover, MODIS Collection 5, and MODIS Cropland. The results showed that the accuracies of cropland area and spatial location of Globe Land 30 were higher than the other four products. The cropland areas of Compared with the statistical data, the best goodness of fit was obtained using Globe Land 30, followed by MODIS Collection 5 and FROM-GLC, with MODIS Cropland and Glob Cover having the poorer accuracies. Regarding the spatial location of cropland, Globe Land 30 achieved the best accuracy, followed by FROM-GLC and MODIS Collection 5, with Glob Cover and MODIS Cropland havi In addition, the spatial agreement of the five datasets was reduced from agricultural production area to pastoral area and significantly affected by elevation and slope factors. Although the spatial resolution of MODIS Collection 5 was the lowest, accuracies of the cropland area and spatial location were better than those of Glob Cover and MODIS Cropland. Therefore, high spatial resolution remote sensing images can help to improve the accuracy of the dataset during land cover mapping while it is also important to select a suitable classification method. Further, in northwestern and southeastern China, spectral mixing pixels are universal because of the complicated landscape and fragmentized topography and result in uncertainty and poor consistency when using the five products. Therefore, these regions require additional attention in future cropland mapping studies.