Biological pest control is an important ecosystem service in agricultural and pastures areas, and can bring both economic and environmental benefits. In areas of tropical pastures, spittlebugs can bring losses of several millions per year, thus being a pest of great importance. To understand the factors that potentially regulate the biological control of these froghoppers, we evaluated through predation experiments the effects of forest cover amount, fragment size and distance in the matrix to fragment on predation process. In addition, we tested the effect of different methodological procedures in the detection of relationships between pest control and landscape structure. We worked with nine landscapes inserted in the Atlantic Forest region, and for each landscape used three transects of 100 m length inside pastures, orthogonally arranged from the forest edge. Predation rate showed a positive relationship with forest cover within 400 m and with fragment size, and a negative relationship with distance from forest fragment. Predation patterns varied with the type of prey used and the exposure time of the experiment. The sentinel prey's experiment (with real preys that are common pests in the region -Deois flavopicta) showed weak patterns with forest cover in 3 km, while the dummy caterpillars experiment (with artificial prey on caterpillars' shape) showed clearer patterns mainly associated with forest cover within 400 m, and with fragment size. The negative effect of distance from the forest fragment became clear only with longer exposure times (96 hours). The predation process in the matrix is thus influenced by the landscape structure, particularly in more local scale (400 m), and different types of prey and exposures times of predation experiments influence the ecological patterns that emerge. Our work highlights the importance of landscape management structure in agricultural areas to optimize the offering of pest control service, as well as the need to use more than one type of experiment to access predation process.