Measuring activity levels in animals provides important information about their behavioral ecology and may be a relevant factor in management and conservation. We tested an individual-based method to discriminate active and passive behaviors on brown bears (Ursus arctos), using a dual-axis motion sensor mounted on Global Positioning System–Global System for Mobile Communications (GPS–GSM) collars. The method takes into account individual variation in activity levels and does not require further calibration. We validated the method through direct observations of captive bears and an extensive survey of wild bear signs in the boreal forest of central Sweden. We found good correspondence between sensor-measured and observed activity on captive bears. Analysis of wild bear signs at GPS locations and its comparison with the collar-based activity status confirmed the possibility of successfully applying the method to study brown bear activity patterns in the wild. The method provided 94.3% correct activity classification on captive bears and about 78.2% on wild bears. We tested the possibility of using this technique to measure increasing levels of activity by analyzing the correlation between the collar-derived numeric activity and the intensity of bear movement. At a broader scale (active vs. passive), the sensor-measured value provided information on the degree of activity, but no correlation was evident at a finer scale (specific behaviors). We suggest that using more sensors in different regions of a bear's body could overcome this difficulty and improve our knowledge of animal behavior in the wild, through remote monitoring of activity levels. We conclude that this method can be useful in the study of behavioral ecology of a wide range of animals, especially species that are difficult to observe or move great distances.