Changes in groundwater storage lead to a reduction in groundwater contribution to river flow and present as non-stationarity, especially during low-flow conditions. Conventional river models typically ignore this non-stationarity, and, hence, their predictions of declines in low flows during drought periods are likely to be compromised. The present study assesses non-stationarity and highlights its implications for river modelling. A quantile regression analysis showed non-stationarity of low flows in the Namoi catchment (Australia), with statistically significant downward trends in the 10th percentile of log-transformed baseflow (10-LTB). This highlighted the usefulness of the 10-LTB metric to identify non-stationarity and, hence, alert modellers to the importance of adopting models that explicitly account for groundwater processes when modelling such river systems.