Streamflow is the major factor influencing the evolution of solute concentration in river water and different modelling approaches exist to characterize the dependency of ion concentration to discharge: the simplest are based on measurable quantities (stream discharge and stream ion concentration) but do not allow for an explicit, physical, flow-path interpretation; the more complex are based on mixing assumptions with different end-members sources, but require the knowledge of (unmeasurable) flow components. We present here a new concentration–discharge model, which associates a classical concentration–discharge relationship with a classical two-component mixing equation. The originality of our approach lies in the fact that we do not proceed in the usual way to perform the hydrograph separation: we use an a priori assumption of the baseflow-quickflow separation to infer the source concentration values, contrarily to the usual (inverse) approach. The other notable originality is that all the parameters of this model depend on the temporal variation of the stream discharge. This combined model was tested on high-frequency ion concentration series from the ORACLE-Orgeval observatory (France). This work demonstrates that high temporal resolution data allows for explicit testing of model performance across different hydrologic scales. Results show that the combined mixing model allows a better estimation of streamflow solute concentration series for most ions tested at inter-annual scale, except for nitrate (which do not exhibit a clear C-Q relationship). Our results also confirm the advantage of coupling a time dynamic hydrological model with static C-Q relations for each of the flow components.