Episodic concentration-discharge (c-Q) plots are a popular tool for interpreting the hydrochemical response of small, forested catchments. Application of the method involves assuming an underlying conceptual model of runoff processes and comparing observed c-Q looping patterns with those predicted by the model. We analyzed and interpreted c-Q plots of acid-neutralizing capacity (ANC) for 133 storms collected over a 7-year period from three catchments in Shenandoah National Park, Virginia. Because of their underlying lithologies the catchments represent a gradient in both hydrologic and geochemical behavior, ranging from a flashy, acidic, poorly buffered catchment to a moderate, neutral, well-buffered catchment. The relative frequency of observed anticlockwise c-Q loops in each catchment decreased along this gradient. Discriminant function analysis indicated that prestorm base flow ANC was an important predictor of loop rotation direction; however, the strength of the predictive relationship decreased along the same gradient. The trends were consistent with several equally plausible three-component mixing models. Uncertainty regarding end-member timing and relative volume and possible time variation in end-member concentrations were key factors precluding identification of a unique model. The inconclusive results obtained on this large data set suggest that identification of underlying runoff mechanisms on the basis of a small number of c-Q plots without additional supporting evidence is likely to be misleading.