The literature has shown that for a single study site and, in some cases, for single watersheds, models can be developed based on Pearson correlations and principal component analysis (PCA). These models can predict the concentrations of desired pollutants based on the concentrations in runoff of more conventional and easy-to-measure parameters. This project aimed to use the methods suggested in the literature to investigate the potential for models to be constructed for pollutants such as metals and organics based on their correlations with the conventional parameters of total suspended solids (TSS), total nitrogen (TN) and total phosphorus (TP). The dataset used in this analysis was the National Stormwater Quality Database Version 3.0 and the Pearson paired correlations that have been finished to date. In general, few paired correlations were found with the traditional surrogates of TSS, TN and TP. This indicates that while site studies or single land use studies may find correlations with these conventional surrogates, transferring these correlations broadly across land uses and geographical regions is more problematic. It is anticipated that models built based on PCA may be able to predict the concentrations of less conventional parameters based on the measured concentrations of several conventional ones.