This paper evaluated 3 Chinese-made satellite images for extraction of land use information in macroscopic monitoring in 6 typical counties with varied topography characteristics. The authors compared many classification methods and analyzed the impacts of training samples, classification methods, and feature data on classification accuracies. It is found that the simple MLC method can acquire better accuracy, whereas the object-oriented method can improve the accuracy for HJ-1 and BJ-1. For CBERS-02B, the MLC classification accuracy tended to be stable and the Kappa coefficient was above 0.8 when the training samples reached 50. For HJ-1 and BJ-1, the classification accuracy tended to be stable when the training samples reached 60 and the Kappa coefficient was above 0.7. The feature data (i.e. NDVI and DEM) can improve the classification accuracy for CBERS-02B, and DEM can improve the accuracy for HJ-1 and BJ-1. The authors suggest integrating multiple Chinese-made satellite images for acquiring valid images all over the country in macroscopic monitoring. The findings provide scientific basis in accuracy and methods for the macroscopic monitoring.