The response of the Mediterranean mean sea level to atmospheric pressure forcing is analyzed using 3 years of TOPEX/POSEIDON data. Coherence analysis between mean sea level and atmospheric pressure shows a significant departure from a standard inverse barometer effect at frequencies higher than 30 days-1. At high frequencies the phase difference between sea level and pressure is about 100¿, while it should be 180¿ for a perfect inverse barometer response. This result is in agreement with previous findings and confirms the role of the Straits of Gibraltar and Sicily in limiting the water exchange (and thus the response to atmospheric pressure forcing) at high frequencies. The response of the Mediterranean mean sea level is then investigated using the Candela <1991> analytical model which takes account of friction in the Straits of Gibraltar and Sicily. The model explains a large part of the variance in TOPEX/POSEIDON mean sea level variations (50% for the western basin and 38% for the eastern basin). Compared to an inverse barometer correction, it gives a smoother response with a phase delay at high frequencies. It also explains more variance in TOPEX/POSEIDON mean sea level variations (5 cm2 and 7 cm2 for the western and eastern basins, respectively). This demonstrates that this simple model provides an improved correction of atmospheric pressure effects in TOPEX/POSEIDON data. As the two corrections have an rms difference of 2--3 cm with maximum differences of up to 10 cm, the impact on the mapping of oceanic circulation is not negligible. This is exemplified through the comparison of sea level anomaly derived from the two corrections.¿ 1997 American Geophysical Union