Abundance estimation is paramount for the conservation and management of mammalian carnivores. The use of camera-traps for the study of carnivores has exponentially increased in the last decades, mostly for density estimation of species with individual natural coat patterns. Since canid species are not easily individually recognized from their pelage design, their population density is rarely estimated from camera-trap data. We assessed camera-trap data to estimate red fox (Vulpes vulpes) abundance in two areas of SW Europe with contrasting densities. We compared estimates from spatial mark-resight (SMR) models in a Bayesian approach, and assessed the effect of individual identifications by different observers, using as reference the density estimate including as identified exclusively those tagged individuals (unequivocally recognized). We also assessed whether individual identification with ear-tags and GPS-collar movement data improved the precision of estimates. Our results show a high variability among observers in the number of individuals identified and, consequently, in the density estimates, ranging between 1.05 and 1.58 foxes km-2 in the high-density area, and between 0.19 and 0.44 foxes km-2 in the low-density area. The density estimates obtained from spotlight-counts and distance sampling methodology are close to estimates from SMR models. Selected SMR models included a sex-effect, consistent with radio-tracking data indicating that males moved over larger ranges than females. The effects of behavior response and different detection due to lure/bait were also included in the models. We discuss the consequences and applications of these results for monitoring, conserving and managing canids with cryptic pelage patterns.