We used open population, spatial capture-recapture (SCR) models to estimate sex-specific density, survival, per capita recruitment, and population growth rate of ocelots (Leopardus pardalis) at five sites in Belize with up to 12 yr of data per site. Open population SCR models enabled us to separate survival and recruitment from migration using an ecologically realistic, spatially explicit movement model. Yearly survival probability across 4 broadleaf forest sites was estimated at 0.73-0.84 for males and 0.81-0.87 for females, with no clear indication of sex differences. Yearly per capita recruitment was estimated across four broadleaf forest sites at 0.06-0.08 recruits/N for males and 0.09-0.12 recruits/N for females, again with no clear indication of sex differences. At a pine forest site with a population comprised largely of males, survival and recruitment estimates were similar to the broadleaf sites. Population densities in the broadleaf forest sites ranged from 6.5 to 14.7 ocelots/100 km(2), and 0.9-2.5 ocelots/100 km(2) in the pine forest site, with strong evidence of a female-biased sex ratio in the broadleaf sites and a male-biased sex ratio in the pine forest site. We also found strong evidence that female within-year space use at the broadleaf sites was smaller than that of males, and that within-year space use at the pine forest site was larger than that at broadleaf sites. Between-year home-range relocation at broadleaf sites was of a similar spatial scale as within-year space use, consistent with philopatry. We found evidence of a small population decline (posterior probability > 0.9) at two of four broadleaf sites; however, given the level of uncertainty about decline magnitudes, we suggest continued monitoring of these sites to increase site-years and gain further precision on population growth rate estimates. Estimating demographic parameters at large spatial and temporal scales is important for establishing reliable baseline estimates for future comparison and for understanding changes in population dynamics. Long-term data sets like those we collected are of particular importance for long-lived species living at low densities and large spatial scales, where not many individuals are exposed to capture in any one year.