Atmospheric data from reanalysis, satellite, and experimental observations have been combined to calculate a four‐year time series of the Atlantic inflow through the Strait of Gibraltar. The net flow through the strait, estimated from the Mediterranean water budget, and the Mediterranean outflow, estimated from currentmeter observations in Espartel sill (western Strait of Gibraltar) from October 2004 to January 2009, made it possible to estimate the Atlantic inflow as the sum of both of them. The obtained mean net flow is 0.038 ± 0.007 Sv, with a seasonal cycle of 0.042 ± 0.018 Sv annual amplitude and maximum in September. The Mediterranean outflow shows a seasonal signal with annual amplitude of 0.027 ± 0.015 Sv peaking in April (in absolute value), and a mean value of −0.78 ± 0.05 Sv. The resulting Atlantic inflow has a mean value of 0.81 ± 0.06 Sv and a seasonal cycle with annual amplitude of 0.034 ± 0.011 Sv, peaking in September, and high interannual variability. The inflow seasonal cycle is the result of a barotropic forcing associated with the cycle of the net flow, driven by the evaporative cycle, and a baroclinic forcing linked to the seasonal cycle of the reduced gravity that drives the exchange.

Three‐yearlong time series of Acoustic Doppler Current Profiler (ADCP) observations at a single station in Espartel Sill (Strait of Gibraltar) were used to compute an outflow of Q₂ = −0.82 Sv through the main channel. The cross‐strait structure of the velocity field or the outflow through a secondary channel north of the submarine ridge of Majuan in Espartel section is not captured by observations so that an improved version of a numerical model (CEPOM) has been used to fill the observational gap. Previously, the model performance has been checked against historical data sets by comparing harmonic constants of the main diurnal and semidiurnal constituents from observed and modeled data at different sites of the strait. Considering the great complexity of tidal dynamics in the area, the comparison is quite satisfactory and validates the model to infer the exchange at longer timescales. Using a “climatological” April in the simulation, extracting a “single station” from the model at the same position as the monitoring station and processing the data similarly, the model gives an outflow through the southern channel 13% higher than observations. The inclusion of the cross‐strait structure of velocity reduces the computed outflow through the southern channel, whereas the contribution of the northern channel brings the total outflow close to that computed using a single station (5% smaller). If the same correction is applied to observations, the total outflow would reduce to Q₂ = −0.78 Sv. The paper also assesses the importance of eddy fluxes to the total outflow, their contribution being negligible (≤5%).