Global Positioning System (GPS) telemetry allows to monitor large herbivores in Alpine habitats that are difficult or impossible to access by humans. However, little is known about the factors influencing the performance of GPS telemetry in these habitats. We used stationary test collars and a data set of over 90000 attempted locations from 11 collared free-ranging Alpine ibex (Capra ibex ibex) females to quantify the loss of scheduled locations and accuracy of GPS telemetry at the high-elevation habitats of the Marmolada massif (eastern Italian Alps). In both tests the fix acquisition rates and proportions of 3D fixes averaged near 90%. After removing outliers, the estimated location error in stationary collar tests was within 10 m for 75% of 3D locations and within 20 m for 75% of 2D locations. In both tests, sky view determined by terrain morphology was the main habitat feature influencing the performance of GPS telemetry, while forest had only minor effects compared to open ground. Both fix acquisition rates and proportions of 3D fixes in free-ranging ibexes were lowest (close to 85%) in winter and highest (close to 95%) in summer, and exhibited contrasting seasonal diel patterns, being lowest during the night and highest at midday in winter, and lowest at midday and highest at dawn and dusk in summer. Fix acquisition rates and proportions of 3D fixes were positively influenced by individual activity, warm temperatures in winter, absence of precipitation in all seasons, and shallow snow depth in winter. These effects could explain the seasonal and diel patterns. The performance of GPS telemetry of large herbivores at Alpine habitats can be as good as or better than in forested habitats at low elevations. However, acquired locations underestimate inactive periods in sheltered habitats, especially in winter and in adverse weather conditions.