As part of Lawrence Berkeley National Laboratory’s national cross-sectional study of individuals who live within 5 miles of a modern, utility-scale wind turbine, 15 wind power projects were selected as case studies and over-sampled. The same 15 wind power projects were modeled to estimate the sound levels at each respondent’s home. Also, a representation of background sound level for each respondent was extracted from a national dataset. Statistical analyses were conducted to estimate the acoustical contributions to one’s propensity for annoyance, and how these were affected by non-acoustic factors (e.g., project compensation, prior attitude toward the project, visibility, etc.). The results demonstrate that considering the interaction of a project’s modeled sound levels and the existing background sound levels improves the prediction of reported wind turbine audibility over only using modeled sound levels. Additionally, the sound-level drivers (modeled wind turbine sound level and background sound level) are poor predictors of very annoyed responses; one’s prior support for or opposition to a local project is the strongest predictor of very annoyed responses in the regression model.