Conservation and management of snow leopards (Uncia uncia) has largely relied on anecdotal evidence and presence–absence data due to their cryptic nature and the difficult terrain they inhabit. These methods generally lack the scientific rigor necessary to accurately estimate population size and monitor trends. We evaluated the use of photography in capture–mark–recapture (CMR) techniques for estimating snow leopard population abundance and density within Hemis National Park, Ladakh, India. We placed infrared camera traps along actively used travel paths, scent-sprayed rocks, and scrape sites within 16- to 30-km² sampling grids in successive winters during January and March 2003–2004. We used head-on, oblique, and side-view camera configurations to obtain snow leopard photographs at varying body orientations. We calculated snow leopard abundance estimates using the program CAPTURE. We obtained a total of 66 and 49 snow leopard captures resulting in 8.91 and 5.63 individuals per 100 trap-nights during 2003 and 2004, respectively. We identified snow leopards based on the distinct pelage patterns located primarily on the forelimbs, flanks, and dorsal surface of the tail. Capture probabilities ranged from 0.33 to 0.67. Density estimates ranged from 8.49 (SE = 0.22) individuals per 100 km² in 2003 to 4.45 (SE = 0.16) in 2004. We believe the density disparity between years is attributable to different trap density and placement rather than to an actual decline in population size. Our results suggest that photographic capture–mark–recapture sampling may be a useful tool for monitoring demographic patterns. However, we believe a larger sample size would be necessary for generating a statistically robust estimate of population density and abundance based on CMR models.