Artificial roosts have been proposed as a tool for augmenting bat populations and catalyzing tropical forest regeneration. In the best case scenario, roosts would attract seed-carrying bats (Family Phyllostomidae) into degraded pastures and form nucleating patches of native vegetation. We tested this scenario by monitoring 48 artificial roosts in pastures and adjacent forest fragments in southern Costa Rica over 2 years. Half of the pasture roosts were exposed to direct sunlight and half were affixed to 4-m living stakes of Erythrina poeppigiana (Walp.) O.F. Cook that provided shade. After 2 years, 94% of roosts in forest and 40% of roosts in pasture had been used by bats at least once – primarily for nocturnal feeding. Maximum daily temperature inside of roosts was the best microclimatic predictor of bat visitation. We identified at least five species of bats that visited roosts, including two frugivores (Carollia and Glossophaga spp.). Bat-mediated seed dispersal increased with the number of frugivorous bat detections at roosts, but seedling recruitment did not increase with either bat detections or seed abundance over a 2-year period. Given that bats rarely used roosts in pastures, and bat visitation did not increase seedling recruitment, our data suggest that artificial bat roosts did not accelerate forest regeneration in abandoned, premontane pastures in southern Costa Rica. This method could be refined by investigating alternative roost designs, barriers to seedling recruitment below roosts, improvement of roost microclimatic conditions in pastures, and ability of bats to detect roosts in different habitats.