The density of colonies of leaf-cutting ants, Atta cephalotes L. (Hymenoptera: Formicidae), was measured and compared among coffee (Coffea arabica L.) plantations in five management categories: monoculture conventional, diversified conventional, diversified organic, highly diversified conventional, and highly diversified organic. Twenty-four small farms (<4 ha) in Turrialba, Costa Rica, were included in this study. Within-farm and off-farm (landscape) variables were measured and tested for their relationship with A. cephalotes colony density. Total ant colony density (colonies per ha) and density of new colonies shortly after a nuptial flight were significantly greater in the coffee monoculture conventional system, compared with all other systems. Total ant colony density and density of new colonies were inversely related to percentage of shade within the farms. Within farms, colony density was greater near edges adjacent to riparian forest than those adjacent to nonforested land. Regardless of edge type, plots closer to the edge (0–30 m) had greater colony density than those furthest from the edge. At the landscape scale, density of new colonies was positively related to fallow land use coverage within a 2,000-m buffer radius and to forest coverage within a 500-m radius. Results indicate that coffee farm management practices and landscape level factors can affect A. cephalotes colony densities. Understanding such practices and factors could assist in the development of better management methods of these injurious insects in coffee farms. Increased diversification in coffee farms, possibly due to the greater shade associated with it, may reduce colonization by the ants, which are considered forest gap specialists.