In terrestrial walking organisms, long legs help to decrease the cost of running, allowing animals to step over environmental interstices rather than walking through them. However, long legs can complicate the infiltration of these interstices, which may contain food sources and refugia. Since the number of environmental interstices perceived by an organism (rugosity) increases as it body size decreases (size-grain hypothesis, SGH), natural selection should favor proportionally smaller legs with decreasing body size. Recent work demonstrated that ants fit this hypothesis. We experimentally tested the assumption of the SGH that small ants, which have proportionally smaller legs than larger ants, are more successful in exploring environmental interstices because they can easily penetrate them. We examined the ability of tropical litter ant species with different body sizes to access food baits in ‘landscapes’ (=plots) with different levels of rugosity and food exposure. In the first experiment, three levels of landscape rugosity were defined by manipulating the density of leaf litter placed on the ground plots: a) plain landscape: no litter fall, b) intermediate rugosity (∼0.5 kg of litter fall) and c) high rugosity (∼1 kg). In a second experiment, food baits were in plain landscapes, exposed or covered by leaf litter. The body lengths of ants that first accessed food baits ranged from 1.5 to 12 mm. Ants that first reached food baits in the most rugose landscapes were ∼40% smaller than ants that first found baits in plain landscapes. Smaller ants were also the first to access covered food. The application of a phylogenetic comparative method suggested the same patterns. We conclude that these results support the size grain hypothesis. Environmental rugosity might have operated as a selective force to shape the morphological characteristics of litter ant species.