Path integration involves updating one's position and orientation based on information about self-motion, such as optic flow or body senses (vestibular information, proprioception, and efference). In the human navigation literature, there is evidence that people can use the body senses for path integration while walking without vision (Loomis, et al 1993) and use optic flow when body senses are unavailable, as in a joystick task (Peruch, et al 1997, Riecke, et al 2000, Bud, 2000). However, when both types of information are available, the body senses appear to dominate (Bud, 2000). To dissociate optic flow and the body senses in a triangle completion task, we manipulated the optic flow rate during active walking in an immersive virtual environment. The visual gain was either greater than (150%), lower than (67%), or the same (100%) as in normal walking, and gains for observer translation and rotation were varied separately. In Exp. 1, the body senses played a strong role, but the influence of optic flow was indeterminate because the visual gain covaried with the physical size of the walked triangle. When we eliminated this covariation in Exp. 2 we confirmed the dominance of body senses. In addition, we found that there is a small but reliable role of optic flow for path integration. We are currently investigating whether the addition of landmarks produces a stronger visual contribution to path integration.