This paper describes an anomaly behavior pattern detection system for elderly people using a temperature sensor. The proposed monitoring system automatically detects anomaly behavior patterns using fuzzy inference. The proposed fuzzy system is constructed by some feature values such as the center of gravity and variance extracted from the human region in two-dimensional temperature array data. The change amount of human region in the past twenty samples is also utilized. Anomaly behavior patterns are detected based on the possibility distribution inference considering the behavior time zone and duration. Some experiments are provided to verify the effectiveness of the proposed system.