1.Avian electrocution on power lines is a major conservation issue on a global scale. Electrocution risk models have recently been proposed as an effective alternative to prioritising high‐risk pole retrofitting activities at a large scale. However, existing models ignore the specific features of the power poles supporting the power distribution lines and make the tenuous assumption that pole density and power line length are key factors to assessing the electrocution risk at a large scale. This assumption may be violated in areas with high variations in pole configuration. 2.In this study, we used data on raptors electrocuted on poles to develop a predictive model of raptor electrocution risk throughout an extensive geographic area in north‐western Spain, using boosted regression trees (BRT). With the best‐fitting model we predicted the hazard of a set of 188 741 poles and validated the model predictions with new data collected from the study area. 3.Our model highlights the relevance of combining both habitat and technical features to identify the most dangerous poles for raptors on a large geographic scale. A 9·86% of the total poles evaluated were characterised as high risk for raptors. The model showed good performance in external validation. The new electrocution events were registered at poles with high‐risk values. 4.Synthesis and applications. In this study we improved the accuracy of the predictive models of raptor electrocution risk for large geographic areas. By incorporating the technical characteristics of the power poles into the models, we achieved a high level of prediction at the power pole level which is the ultimate correction unit. This will allow electric companies and wildlife managers to specify retrofitting activities of high‐risk power poles for raptors in large geographic areas, thus maximizing the effect of investment in the correction of dangerous power poles and conservation of the raptor populations.