Submarine Groundwater Discharge (SGD) has received increased attention in recent years since it was recognized that it may be both volumetrically and chemically important. Around the Mediterranean Sea, 60% of the coastline is composed of karstic aquifers, and to properly estimate the hydrological budget of the Mediterranean Sea it is therefore necessary to better assess the karstic submarine groundwater discharges (KSGD). However, quantifying KSGD is still challenging. Among the methods recently developed to detect and quantify SGD, the mass balance method of the radium quartet 223 Ra, 224 Ra, 226 Ra and 228 Ra has proved to be a powerful technique. This approach requires characterizing all the contributing terms and sinks in the coastal water volume affected by SGD, the residence time of coastal waters, as well as a representative concentration of the tracers for both surface water and discharging groundwater. In this study we combine several approaches (223 Ra, 224 Ra, salinity profiles and Acoustic Doppler Current Profiler (ADCP) measurements) to examine both the accuracy and sensitivity of the radium mass balance method in the case of the cove of Port-Miou (Mediterranean Sea, France) where the main karstic spring discharges locally at 10 m depth. This study benefits from the inland in-situ access to the main karst conduit discharging to the sea which provides a long time series to characterize 2 the brackish submarine groundwater end-member. We show that the composition of the cove water is stratified, with two water bodies: a surface brackish layer and a deeper layer. The mean KSGD value obtained with 223 Ra and 224 Ra mass balances in the surface water body is precise but significantly lower (0.6 ± 0.1 m 3 /s) than the karstic spring discharge (4 ± 1 m 3 /s) estimated within the karst conduit with pressure sensors. The residence time of the cove water estimated using both 224 Ra and 223 Ra isotopes is very low (1 ± 1 day). Our study shows that the water residence time that we calculated using the Ra mass balance is the key parameter that may impact KSGD. In addition, based on ADCP transects, we suggest that the shape and geometry of the cove, as well as the location of the discharge point of the spring play a key role in explaining these discrepant results. We therefore recommend that in such stratified coves, estimations of KSGD based on short-lived radium isotopes require accurate and independent estimates of the water residence time as well as a good knowledge of the shallow and deep circulation patterns of the cove water.