American pikas (Ochotona princeps) have been heralded as indicators of montane-mammal response to contemporary climate change. Pikas no longer occupy the driest and lowest-elevation sites in numerous parts of their geographic range. Conversely, pikas have exhibited higher rates of occupancy and persistence in Rocky Mountain and Sierra Nevada montane ‘mainlands’. Research and monitoring efforts on pikas across the western USA have collectively shown the nuance and complexity with which climate will often act on species in diverse topographic and climatic contexts. However, to date no studies have investigated habitat, distribution, and abundance of pikas across hundreds of sites within a remote wilderness area. Additionally, relatively little is known about whether climate acts most strongly on pikas through direct or indirect (e.g., vegetation-mediated) mechanisms. During 2007–2009, we collectively hiked >16,000 km throughout the 410,077-ha Glacier National Park, Montana, USA, in an effort to identify topographic, microrefugial, and vegetative characteristics predictive of pika abundance. We identified 411 apparently pika-suitable habitat patches with binoculars (in situ), and surveyed 314 of them for pika signs. Ranking of alternative logistic-regression models based on AICc scores revealed that short-term pika abundances were positively associated with intermediate elevations, greater cover of mosses, and taller forbs, and decreased each year, for a total decline of 68% during the three-year study; whereas longer-term abundances were associated only with static variables (longitude, elevation, gradient) and were lower on north-facing slopes. Earlier Julian date and time of day of the survey (i.e., midday vs. not) were associated with lower observed pika abundance. We recommend that wildlife monitoring account for this seasonal and diel variation when surveying pikas. Broad-scale information on status and abundance determinants of montane mammals, especially for remote protected areas, is crucial for land and wildlife-resource managers trying to anticipate mammalian responses to climate change.