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.