Understanding the dynamics of flow and recharge processes is crucial information for managing waterresources. However, this is challenging in heterogeneous systems, e.g. karst, due to the strong spatialvariability in flow pathways. In addition, the evolution of mixing at the event-scale is rarelycharacterized with sufficient details to constrain or validate hydrodynamic models. The objectives ofthis study were to (1) identify event-scale variations in contributions from different flow pathways to akarst spring, and (2) better constrain the response of spring dynamics to the spatio-temporal variabilityof flood events. The study focuses on the main spring (Lez spring) of a Mediterranean karst aquifernear Montpellier (SE France), which is subject to intense autumn rainfall events. We used innovationsin (1) the continuous and high frequency monitoring of tracers (delta18O, delta2H, and naturalfluorescence+ humic and proteic-like compounds), and (2) the types of tracers monitored (natural andanthropogenic dissolved gases, radon and radium isotopes, 3D-fluorescence and total organiccarbon). Three rainfall events highlighted distinct spring chemical responses. The first event wascharacterized by a contribution of deep groundwater flow (low CFC contents, low (228Ra 226Ra)ratios and low organic carbon concentrations) resulting from a piston flow effect due to heavy rainfallsat the basin margins. The second event showed a proteic peak associated with a sharp decline in DOconcentrations and an anomalous peak in delta18O. The third event was characterized by a dilutioneffect by subsurface waters (CFC contamination, high turbidity and radon peak) due to high localrainfall. The simultaneous monitoring of these tracers during multiple rainfall events provides aninnovative approach in spring dynamics analysis, and improves our understanding of a complexhydrogeological system. The Lez spring supplies drinking water for Montpellier, thereforeunderstanding the variability of water origins and potential contamination pathways during intensestorm events is essential.