Summary This paper describes a model for simulating flocks of agents by simulating some of the general principles that govern an individual agent's behavior within a flock or group. The goal of this paper is to present a system that can realistically animate the flocking behavior of a group of birds or a school of fish. Key Ideas * The simulated flock is an elaboration on a particle system where each agent is represented by a particle. * There are three main behavioral principles that govern flocking: collision avoidance, velocity matching, and flock centering. An agent should try to avoid collisions with nearby flockmates, attempt to match the velocity with nearby flockmates, and attempt to stay close to nearby flockmates. These behavioral "urges" are expressed as acceleration requests. * Flocking actually depends upon each agent having a limited, localized view of the world. For this reason, an agent's perception model is very ad hoc and avoids higher-level input devices such as simulating vision.