Use of photographic capture–recapture analyses to estimate abundance of species with distinctive natural marks has become an important tool for monitoring rare or cryptic species, or both. Two different methods are available to estimate density: nonspatial capture–recapture models where the trap polygon is buffered with the half or full mean maximum distance moved by animals captured at more than 1 trap (1/2 MMDM or MMDM, respectively); or spatial capture–recapture (SCR) models that explicitly incorporate movement into the model. We used data from radiotracked Eurasian lynx (Lynx lynx) in the northwestern Swiss Alps (NWSA) during a low (1.0 lynx/100 km²) and a high (1.9–2.1 lynx/100 km²) lynx population density to test if lynx space use was density dependent. Second, we compared lynx density estimates resulting from these 2 different methods using camera-trapping data collected during winters 2007–2008 and 2009–2010 in the NWSA. Our results indicated lynx space use was negatively correlated with density. Lynx density estimates in all habitats using MMDM (0.86 and 0.97 lynx/100 km² in winters 2007–2008 and 2009–2010, respectively) were significantly lower than SCR model estimates, whereas there was no significant difference between SCR model (1.47 and 1.38) and 1/2 MMDM (1.37 and 1.51) density estimates. In the NWSA, which currently harbors the most abundant lynx population in Switzerland, 1/2 MMDM and SCR models provided more realistic lynx density estimates compared to the MMDM, which lies in the lower range of densities. Overall, the SCR model is preferable because it considers animal movements explicitly and is not biased by an informal estimation of the effective sampling area.