The aim of this study is to develop a prediction model to estimate the fatigue levels of individuals using physiological measurements. Background: Advances in wearable sensors have opened new opportunities to measure real-time physiological states of individuals. By exploiting such technologies, it is possible to monitor a variety of cognitive states of a user using measured physiological states. Method: We measured acceleration and photoplenthysmogram (PPG) using a wearable sensor at the same time while participants performed physical exercises. Physical exercises included running, walking and weight training programs for hours. Participants reported subjective levels of fatigue after exercise programs. A regression based model was developed to predict the subjective fatigue level from acceleration and PPG data. Results: The developed model could predict subjectively perceived fatigue levels from acceleration and PPG signals with accuracy higher than a chance level. Conclusion: The present study demonstrates a feasibility of predicting individual fatigue levels using physiological data measured by a commercially available wearable sensor. Application: The prediction model can be applied to home automation systems to adaptively adjust home environments according to the user’s fatigue states.