The sea level output from two barotropic models, Hindcast of Dynamic Processes of the Ocean and Coastal Areas of Europe (HIPOCAS) and Dynamic Atmospheric Correction (DAC), are compared and evaluated on the basis of coastal tide gauges (TG) and TOPEX/POSEIDON measurements in the Mediterranean Sea and the NE Atlantic Ocean for the period 1993–2001. Both models reduce the observed sea level variance more than the classical inverted barometer correction. However, differences between the models arise for different regions and frequency bands. In coastal areas, Hindcast of Dynamic Processes of the Ocean and Coastal Areas of Europe (HIPOCAS) reproduces observed (TG recorded) sea level better than DAC (residual variance of 70.81 ± 0.69 cm2 versus 74.05 ± 0.68 cm2). This is particularly true in the Atlantic Iberian coasts (49.58 ± 1.09 cm2 versus 68.53 ± 1.12 cm2), where HIPOCAS is able to reproduce a wind‐generated signal probably linked with seasonal upwelling. The exception is the northern Adriatic, where HIPOCAS gives higher residual variance than DAC (118.80 ± 0.60 cm2 versus 107.15 ± 0.60 cm2). At low frequencies (T > 20 days) the atmospherically induced coastal sea level is better reproduced by HIPOCAS in the entire domain (23.43 ± 0.34 cm2 versus 32.35 ± 0.33 cm2). At high frequencies (T < 20 days), DAC and HIPOCAS perform on average similarly (37 ± 0.5 cm2). In the open ocean, both corrections provide equivalent results (60 ± 5 cm2 residual altimeter variance). Our general recommendation would be to use either DAC or HIPOCAS for the correction of altimetry, and to use HIPOCAS for coastal studies aiming at separating the atmospheric contribution to sea level variability from the steric and mass contributions.