Animal locations collected by Global Positioning System (GPS) collars will represent a biased sample of the sites an animal used if some position fixes fail and if those missed locations do not occur randomly. Probability of a GPS receiver obtaining a position fix is known to decline as canopy cover increases, but the impact of forest canopy cover was insufficient to account for the low fix rates we observed for GPS collars on grizzly bears (Ursus arctos). We tested the hypothesis that GPS fix rates were related to the interaction between animal activity (active vs. resting) and canopy cover by evaluating the following predictions: 1) grizzly bear activity should follow a circadian pattern similar to the circadian fix-rate pattern, 2) grizzly bear use of canopy cover should follow a circadian pattern similar to the circadian fix rates, 3) grizzly bear activity should be related to canopy cover (i.e., bears should rest in areas with relatively high canopy covers and feed and move in relatively open areas), and 4) collar orientation and canopy cover should interact to affect the fix rates of test collars. The GPS fix rates traced a bimodal circadian pattern that was directly related to the circadian pattern of grizzly bear activity. Fix rates declined when bears were more likely to be using denser cover, and fix rates of test collars demonstrated that collar orientation interacted with canopy cover, such that fix rates declined much more with increasing canopy cover when the collar was on its side than when the collar was upright. We concluded that inferences made about grizzly bear microhabitat use, based on GPS locations, will underrepresent high canopy cover sites, especially when grizzly bears are resting there.