A fully coupled regional climate system model (CNRM-RCSM4) dedicated to the Mediterranean region is described and evaluated using a multidecadal hindcast simulation (1980–2012) driven by global atmosphere and ocean reanalysis. CNRM-RCSM4 includes the regional representation of the atmosphere (ALADIN-Climate model), land surface (ISBA model), rivers (TRIP model) and the ocean (NEMOMED8 model), with a daily coupling by the OASIS coupler. This model aims to reproduce the regional climate system with as few constraints as possible: there is no surface salinity, temperature relaxation, or flux correction; the Black Sea budget is parameterised and river runoffs (except for the Nile) are fully coupled. The atmospheric component of CNRM-RCSM4 is evaluated in a companion paper; here, we focus on the air–sea fluxes, river discharges, surface ocean characteristics, deep water formation phenomena and the Mediterranean thermohaline circulation. Long-term stability, mean seasonal cycle, interannual variability and decadal trends are evaluated using basin-scale climatologies and in-situ measurements when available. We demonstrate that the simulation shows overall good behaviour in agreement with state-of-the-art Mediterranean RCSMs. An overestimation of the shortwave radiation and latent heat loss as well as a cold Sea Surface Temperature (SST) bias and a slight trend in the bottom layers are the primary current deficiencies. Further, CNRM-RCSM4 shows high skill in reproducing the interannual to decadal variability for air–sea fluxes, river runoffs, sea surface temperature and salinity as well as open-sea deep convection, including a realistic simulation of the Eastern Mediterranean Transient. We conclude that CNRM-RCSM4 is a mature modelling tool allowing the climate variability of the Mediterranean regional climate system to be studied and understood. It is used in hindcast and scenario modes in the HyMeX and Med-CORDEX programs. Keywords: Mediterranean Sea, regional climate system model, air-sea fluxes, hindcast simulation, interannual variability (Published: 18 November 2014) Citation: Tellus A 2014, 66 , 23967, http://dx.doi.org/10.3402/tellusa.v66.23967