 This study compares the Maxent model with other machine learning models to explore its advantages in simulating urban growth and interpreting driving mechanisms using land use data from Global and 30. The results indicate that the Maxent model outperforms the other models except for computational efficiency, but the time required for Maxent training and projecting is acceptable and less than that of SVM. The Maxent CA model reveals complex driving mechanisms of urban growth in Tianjin due to more variable interactions while the relationships between spatial factors and urban growth are non-linear in all three study areas. Topographic factors, city center, traffic factors, and water bodies are significant factors affecting urban growth in Beijing, Tianjin, and Wuhan, respectively. This article was authored by Bin Zhong and Haijun Wang.