Tracking animal movement using Global Positioning System (GPS) technology is an increasingly popular method for studying animal ecology, behavior, and conservation. To date, most GPS location schedules have been set at regular intervals. If intervals are too long, they undersample the details of movement paths, and if too short, they oversample resting sites and deplete the unit's battery without providing new information. We address this problem by creating a dynamic GPS schedule that is linked to the activity level of the animal via an accelerometer onboard the tracking tag. We deployed traditional and accelerometer-informed GPS tags on northern tamanduas anteater (Tamandua mexicana) in tropical forest in the Republic of Panama (2009–2010), and on fisher (Martes pennanti) in temperate forest in New York, USA (2009–2011). These species are medium-sized forest-dwellers that frequently use tree cavities, ground burrows, and thick vegetation for resting and foraging, all traits that make them particularly challenging for GPS tracking. The accelerometer-informed tags performed better than the traditional GPS tags: they attempted 73.6% more locations per day, achieved 61.7% higher location success rates, spent 28.2% less time searching for satellites, made 67.4% fewer redundant location attempts in places where animals were inactive, and ultimately provided more data for a given battery size. The resulting tracks of animal movement had high temporal resolution, revealing aspects of their behavior and ecology that would have been missed by traditional tags, especially for the fast-moving fisher. By dynamically linking the location schedule to animal movement rate, accelerometer-informed GPS tags reduce the trade-off between collecting detailed movement data and recording movement data for a longer period of time. © 2012 The Wildlife Society.