In most states, the phosphorus (P) index (PI) is the adopted strategy for assessing a field’s vulnerability to P loss when preparing comprehensive nutrient management plans. Most state PIs, however, have not been rigorously evaluated against measured P loss data to determine how well the PI assigns P loss risk – a major reason being the lack of field data available for such an analysis. Here we demonstrate an approach for evaluating and revising a PI using P loss data generated by a process-based model. Our first objective was to use regression analysis and model-generated P loss data to evaluate two different index structures, three different runoff factors for assessing P loss risk from surface applied manures and fertilizers, and two different approaches for relating sediment P loss to soil test P. Our second objective was to demonstrate how output from a process-based model can be used to evaluate and modify P index weights. The practical importance of our findings was assessed by comparing how well the different index formulations were able to predict outcomes of existing runoff studies. Our results show that a component formulation provides better correlation to simulated P loss data than a multiplicative formulation, though when evaluated against field-measured P loss, differences between the two formulations were minimal. Applying our approach to an archetypical PI based on the PA P Index significantly improved the correlation between the index and field-measured P loss data. The approach we use here can be used with any P loss model and PI and should serve as a guide to assist states in evaluating the accuracy and making modifications to their PI if necessary.