Populations of the Central American spiny rat Proechimys semispinosus isolated on eight small islands in Panama were sampled monthly for 1 year to examine the relationships of demography to tree species composition and forest structure. Demography was characterized by calculating seven variables from the monthly census data. Tree species composition was determined by marking, measuring and identifying all trees ð 10 cm in diameter at breast height (dbh) on each island. Forest structure (tree stature, density and species richness) was quantified by measuring seven variables at each of the 470 sampling stations. Bivariate correlations were computed between spiny rat demography and tree species composition and forest structure. To examine whether islands with more similar tree species composition or forest structure supported P. semispinosus populations with more similar demography, dendrograms based on complete-linkage cluster analysis were constructed. Spiny rat densities were correlated positively with the density of large-fruited fig trees and non-linearly to an index of forest age based on tree species composition. Spiny rat densities were not related to any structural variable. Dendrograms based on cluster analysis showed that spiny rat demography did not map onto forest structure, overall tree species composition or any group of selected tree species except figs Ficus spp. Thus, islands with more similar fig tree composition contained spiny rat populations with more similar demography. Tree species composition, which partly determines resource abundance, is apparently more important than forest structure in influencing spiny rat demography. Proechimys semispinosus has a temporally and spatially variable demography, and this flexible demography, in concert with an ability to use diverse resources, presumably promotes abundance and persistence in heterogeneous tropical forests and may contribute to the ubiquity of this rodent.