Land use and land cover change is fundamentally important for studying global environmental change and sustainability because it has great impacts on biodiversity, ecosystem processes, atmospheric circulation, environmental quality, climate change, etc. Simulation modeling has been important in the study of land use dynamics and, recently, agent-based models (ABMs) have been developed and applied in this field. An ABM is composed of a set of agents interacting in a common environment. Agents have the following characteristics: goal-directed, autonomy, social abilities, reactivity and pro-activities. They can interact with one another and with the environment and complete tasks autonomously without human intervention. They exhibit goal-directed behavior, have their own problem-solving capabilities, and are able to interact in order to reach an overall goal. Agents perceive, produce, transform and manipulate objects in the environment and reach autonomy cooperatively by sensing activity and modifying their environment. ABMs have an intuitive appeal for simulating complex human-involved processes. They are able to explicitly model individual decision makers and their interactions and simulate complex human behavior and decision processes as they are affected by socioeconomic factors. In urban studies, ABMs have been integrated with human behavior and decisions with cellular automata architecture. In ecology, individual-based models (IBMs) were developed at the end of 1980s which are similar to ABMs in several ways. Now ABMs have been used in economics, sociology, psychology, finance, urban simulation, and other fields. Major platforms include SWARM, RePast, Ascape, and CORMAS. In this paper, we attempt to outline the basic concepts and general modeling framework of ABMs, discuss the pros and cons of these models, and explore their future directions in simulating land use change, urban dynamics, and ecological processes.