We developed a predictive model and map characterizing human-Asiatic black bear contact in Yamanashi Prefecture, central Japan. Our goal was to understand the relationships between human-bear contact and landscape-level physiographic and human variables and to identify potential areas of high risk for contact. Models for years of normal contact and years of recurrent contact were not significantly different. Consequently, the areas of high contact risk were fixed. An integrated model using both normal- and recurrent-year data indicated that the factors affecting contact were, in descending order: distance from forest edge, ratio of conifers to total forested area, bear distribution, and distance from settlement. The classification accuracy of the predictive model was 82.0% (sensitivity 82.0%, specificity 82.0%). This highly accurate predictive map and model may serve as effective tools for reducing human-bear encounters.