The tawny deermouse (Peromyscus perfulvus), a semiarboreal endemic in western Mexico, exhibits attributes associated with extinction vulnerability. In January 2003 and 2004, we conducted mark-and-recapture studies at Playa de Oro, Colima, Mexico. Each year, five 100-station live-trap grids (each station with one ground and one arboreal trap; 10 by 10 configuration) were set up for 8 nights in areas with dense vegetation. Only one tawny deermouse was caught in 2003. In 2004, we trapped 16 individuals (12 males, 4 females) 47 times, with 91.4% of captures being in arboreal traps. Captures in 2004 were at 25 stations, all on the same grid, in habitat representing a mixture of thorn forest, mangrove, and palm trees (density 13.6 mice/ha). Considering all 500 trap stations in 2004, we used stepwise logistic regression to evaluate differences based on 13 structural-vegetation measures of the 25 stations where tawny deermice were found versus the 475 where it was not caught. Tawny deermice tended to inhabit locations with trees close-by, sparse low-level vegetation, little litter, and dense high-level vegetation. For the one grid where we caught tawny deermice (25 stations where present versus 75 absent), only distance to nearest tree was significantly predictive; mice avoided even small forest openings. Tawny deermice were trapped at only a small subset of sites within what heretofore has been considered its suitable habitat. Evidence of restricted spatial structure and other limiting demographic features indicate a need for increased concern in conservation of this endemic species.