Despite the huge diversity tropical arthropods represent, factors shaping their communities are still poorly known, especially at small spatial scales. In this study, we aimed at providing ecological and methodological insights from a short and intensive field sampling of spiders, a highly diverse group of predators. We investigated how sampling methods, habitat type and day-time affect diversity and composition of spider assemblages. The standardized sampling protocol was applied in a tropical rainforest in French Guiana, where both ground- and vegetation-dwelling (up to 2.5 meters) assemblages were sampled during day and night using supposedly complementary methods at low and high (granitic hills called inselbergs) elevations. Observed and estimated richness of vegetation-dwelling spiders, as well as their species composition, did not differ between methods (sweep netting vs. beating). Species richness was much lower in pitfall traps than in litter samples, which suggests a low mobility of ground-dwelling spiders and reveals the inadequacy of the former method compared to the latter. Spider assemblage in the vegetation of inselberg was two times poorer than in lowland forests and dominated by different families, probably due to harsher habitat conditions there. Strongly different patterns were here highlighted between vegetation and ground-dwelling spider assemblages, the latter being less diversified than in the vegetation which deserves further attention.