Development of water sources for wildlife is a widespread management practice with a long history; however, needs of wildlife and availability of water depend on myriad interacting factors that vary among species and localities. Benefits are therefore situational, establishing a need for evaluation of water use in varied settings. We used global-positioning-system (GPS) collars and time-lapse videography to estimate the distribution of elk (Cervus elaphus) activity and frequency of water-development use at Theodore Roosevelt National Park, North Dakota, during June–September, 2003–2006. Elk were located further than expected from the Little Missouri River and did not preferentially use areas near developments. Of 26,081 relocations obtained at 7-h intervals, 88% were >800 m and 74% were >1600 m from permanent surface water. Elk were videotaped at water developments on 90 occasions during 19,402 h of monitoring but used water in only 52% of cases (SE = 5.3%). The probability of detecting elk at developments during visits was 0.51 (SE = 0.08). Nevertheless, elk tracked with GPS collars at 15-min intervals approached to within 100 m of developments on only 2.7% (SE = 0.6%) of 766 days, and approached randomly selected locations nearly as frequently (x¯ = 2.2%, SE = 0.13%). Our results do not rule out use of drinking water by elk at THRO; however, elk were not dependent on water from developments or the Little Missouri River. Prevailing perceptions of water use by elk derive primarily from general associations of elk activity with locations of water sources. Technological advances that permit nearly continuous, precise monitoring present an opportunity to improve understanding of water use by elk, incidental to other investigations.