Cancer is complex, working at multiple spatial and temporal scales, underpinned by a convoluted network of cell cycle controls and cell-cell interactions. Heterogeneity of cells within tumors has emerged as a key factor in tumor evolution and progression, further adding to the challenge.  After introducing some of the complexity we face, I present four modeling approaches: ecological, cell biology, phenotypic evolution, and network evolution.  I conclude with some of the mathematical and practical challenges we face in better understanding and controlling cancer initiation and progression.