Abstract Cancer behaves as a complex, dynamic, adaptive and self-organizing system, and agent-based models are capable of describing such a system as a collection of autonomous decision-making entities called agents. This review provides an overview of how an agent-based approach can be established, and is being used to model a variety of cancer-related processes including tumor genesis, tumor growth, apoptosis, angiogenesis, vascularization and anti-cancer therapy and discuss both challenges and future directions for agent-based modeling in the field of cancer research. We provide rationales for using holonic agent-based modeling toward the goal of creating realistic simulations of cancer in future research directions. Holonical systems guarantee to provide a recursive and hierarchical modeling for complex systems with dynamic and runtime reorganization. They are adopted for cancer modeling since living organisms have a hierarchical structure and can be decomposed into individual cooperating entities.