We assessed the relative influence of plot and landscape scale factors on coffee berry borer abundance using a variation partitioning approach. We used data from 50 coffee plots, located in Turrialba, Costa Rica, which were characterized for their management, environmental, plant status and landscape attributes. Our results indicated that the number of remaining infested coffee fruits per tree after harvest was the most important variable with 22 % of the coffee berry borer abundance variation explained, followed by the number of fruiting nodes (16 %), altitude (associated to temperature, 13 %), and in a lesser degree, shade type (11 %), spatial dependence between plots (9 %) and proportion of pasture in a 400 m radius (7 %). Our results indicate that picking the remaining infested coffee fruits after harvest is the most important cropping practice for the control of coffee berry borer. However, control can be maximized by integrating plot and landscape scale management activities.