Riverine nutrient export represents a transfer of terrestrial nutrients to lakes, estuaries and the near-coastal zone. In this study, we constructed regional predictive models for riverine dissolved organic carbon (DOC) and organic nitrogen (DON) exports. We used a subset of 10 watersheds to construct regional empirical models of DOC and DON export, reserving two watersheds for testing the predictive ability of each model. For the subset of 10 watersheds, mean watershed soil column C:N ratio explained 75% of the variance in DOC export and 73% of the variance in DON export (p < 0.01). Organic C:N explained 63% and 71% of the variance in DOC and DON exports, respectively. There was a stronger relationship between riverine DOC:DON ratio and mineral soil C:N (R2 = 0.77 p < 0.001) than with organic C:N (R2 = 0.49 p < 0.05), suggesting that de-coupling of DOC and DON dynamics in rivers may occur when hydrologic flow paths favor organic layers. We suggest that mean watershed soil C:N ratio is likely to be an integrator of several controls on riverine DOC export including temperature and precipitation (climatic control), soil texture and nutrient status (edaphic control), vegetative species and their associated micro-flora (biological control) and watershed topography (topographical control). Soil C:N appears to be a useful tool for predicting variability in both DOC and DON flux at a regional scale.