Many processes lead to variability of catchment concentration-discharge relationships, but exports of geogenic (weathering derived) solutes and nutrients (nitrogen and phosphorus species) from agricultural basins display relatively constant concentrations despite large variations in streamflow. These “chemostatic” responses are hypothesized to arise when a large mass store, the parent material for geogenics or chemically recalcitrant legacies of fertilization in agricultural catchments, buffers concentration variability. This hypothesis implies that (1) chemostatic behavior should be a general response to elevated external inputs to a catchment and (2) chemostatic behavior should be predictable from theory. Data- and model-based analyses were used to explore these hypotheses. We evaluated concentration variability relative to discharge (expressed as the ratio of the coefficients of variation of concentration and flow, or CVC/CVQ) across a gradient of increasing exported load, as a proxy for an external impact gradient. The CVC/CVQ of multiple solutes declined with increasing exported load. Exceptions included the geogenic solutes, which showed chemostatic responses for all sites, phosphorus, and some nitrogen species. Nitrate showed a suggestive pattern in CVC/CVQ with export, but further data are needed to confirm its generality. A simple model of runoff generation and solute export suggested that the decline in CVC/CVQ arises if the internal mass store is distributed homogeneously in space and there is sufficient time for mass transfer to reach steady state between runoff events. Export from catchments may become more predictable in impacted watersheds, simplifying water quality prediction but inducing strong hysteresis in recovery and making restoration efforts challenging.