We used remotely triggered cameras to collect data on Puma (Puma concolor) abundance and occupancy in an area of tropical forest in Brazil where the species' status is poorly known. To evaluate factors influencing puma occupancy we used data from 5 sampling campaigns in 3 consecutive years (2005 to 2007) and 2 seasons (wet and dry), at a state park and a private forest reserve. We estimated puma numbers and density for the 2007 sampling data by developing a standardized individual identification method. We based individual identification on 1) time-stable parameters (SP; physical features that do not change over time), and 2) time-variable parameters (VP; marks that could change over time such as scars and botfly marks). Following individual identification we established a capture–recapture history and analyzed it using closed population capture–mark–recapture models. Puma capture probability was influenced by camera placement (roads vs. trails), sampling year, and prey richness. Puma occupancy was positively associated with species richness and there was a correlation between relative puma and jaguar (Panthera onca) abundance. Identifications enabled us to generate 8 VP histories for each photographed flank, corresponding to 8 individuals. We estimated the sampled population at 9 pumas (SE  =  1.03, 95% CI  =  8–10 individuals) translating to a density of 3.40 pumas/100 km². Information collected using camera-traps can effectively be used to assess puma population size in tropical forests. As habitat progressively disappears and South American felines become more vulnerable, our results support the critical importance of private forest reserves for conservation.