The aim of this study is to characterize karst infiltration processes during flood events using a rainfall-discharge model coupled with a transport model for non-conservative solutes. The modelling approach is based on a classical karst devoted model with three connected reservoirs: SOIL (and epikarst) that feeds the aquifer partitioned into DIFFUSE and CONDUIT. Solute transport is modelled using mixing equations, including an empirical retardation factor in SOIL, as well as a first order solute decay. In order to take into account some specificities commonly observed in karst systems, three parameters are added to simulate overflows, piston-type flows in conduits, and a variation of the recharge area according to the initial hydrological conditions. To validate our approach, we used the dissolved organic carbon (DOC) as tracer, which is a non- conservative compound derived from the enrichment of infiltrated water into soil humic substances. The model was applied on a small karst system at Fertans in the French Jura mountains, where discharge and continuous measurements of DOC fluorescence are recorded. The model was calibrated and validated on a set of 19 flood events, showing that the model adequately simulated hydrographs and delayed chemographs during flood events with various rainfall intensities. A large variability of the recharge area was highlighted according to low and high groundwater level periods, and was attributed to the state of hydraulic connectivity in the unsaturated zone. The model simulate the contributions of pre-event and event waters during flood events and allow to better quantify the available resource considering the mixing effect of DOC. It shows, in particular, that total discharge of some flood events during dry periods are mainly composed by pre-event water via piston flow-type processes. Finally, this study show the ability of mixing model to properly simulate solute transport taking into account degradation and retardation processes.