A method of assessing rain forest structure by ranking relative abundance of 41 habitat variables was used to describe habitat differences among six trapping sites (324 subsites). Variables included aspects of all vegetation layers but concentrated on those considered to be important to small mammal distribution. Ordination and classification methods resulted in similar analyses of the data. Differences in habitat structure were primarily related to the moisture conditions of the trapping sites and secondarily to their successional age (regenerating versus primary forest). The most important habitat variables for differentiating between sites were LITTER, CLEARING, SOILS, PIG DAMAGE, FLOODING, FAN PALM, EMERGENTS, CANOPY SURFACE and SEEDLINGS. Habitat structure also varied within sites with some suggestion of small-scale patterning.Small mammal captures were more likely in drier sites and subsites, but there was no difference in trap success between regenerating sites and primary forest sites. Greater numbers of species were captured in sites containing a variety of habitats, a discrete layering of vegetation and an extensive understorey. Small mammal captures were positively associated with five habitat variables (EMERGENTS, LITTER, ROTTING LOGS, SEEDLINGS, ROUGH BARK) and negatively associated with five others (LAYERS, BERTAM, SEDGES, PIG DAMAGE, FLOODING). The destruction caused by pigs is thought to be a major factor since it reduces litter and food availability over wide areas. Leopoldamys sabanus was the most abundant small mammal captured (40/68 individuals) and trap success differences among sites (0.4–1.9%) reflect its preference for higher, well-drained habitats.The study demonstrates the usefulness of a simple method of ranking habitat features according to importance/abundance thus eliminating the all but impossible task of direct measurements in this complex system. This simple method of habitat description provides a basis for studying variables influencing faunal distribution patterns.