Vertebrate population densities were quantified in lowland central Guyana using line-transect censuses at three forest sites subjected to reduced-impact logging (RIL), and three adjacent unlogged sites. We censused a range of forest vertebrate species including large canopy-dwelling and terrestrial birds, three primates, one rodent and one tortoise. Two 4 km transects at each site were repeatedly surveyed during the wet season of 2008 to derive population density estimates on the basis of a cumulative survey effort of 416 km. RIL had ended within 16 months, and sites had been subjected to a mean extraction rate of 3.9 m3 ha-1, equivalent to only 1.1 trees ha-1. Three of the 15 vertebrate species examined here exhibited significantly different abundances at forest RIL sites, two of which were negative. Large frugivores such as primates were less abundant in sites subject to RIL, whereas smaller frugivores, granivores, folivores and insectivores were more common in logged sites. We are unable to reliably distinguish between responses of different taxonomic groups, since robust abundance metrics could only be estimated for four mammal species. Despite this, species traits including dietary guild, body mass, home range size and vertical stratification of forest use are used to explain varying responses. Our findings suggest that responsible reduced-impact logging practices in neotropical forests can be considered as a relatively benign form of forest management that can coexist with the requirements of both local economies and biodiversity conservation. However, our study sites experienced comparably low extraction rates, and detrimental effects such as hunting were low. Our results therefore provide an opportunity to scrutinise the effects of best practice logging systems, though do not necessarily represent typical circumstances across tropical forests.