Sea breeze dynamics in the Marseille area, in southern France, is investigated in the framework of the ESCOMPTE experiment (field experiment to constrain models of atmospheric pollution and emissions transport; Cros et al., 2004) conducted during summer 2001 in order to evaluate the role of thermal circulations on pollutant transport and ventilation. The topography of the surroundings is complex (see Fig.1 et 2) with the presence of the Mediterranean sea and two mountain barriers of different heights separated by the Rhone valley, a gap of 200 km long and 60 km width. This topography suggests that processes described by several authors (for example, Lu and Turco, 1994), inducing both sea breeze circulations, slope flows and valley flows are likely to occur and interact together to generate different air polluted layers in a wide domain around Marseille. Anymore, Fig.1 shows that the coastal shape is not linear and becomes a regional feature to be accounted for in the alteration of the sea breeze (Gilliam et al., 2003). The aim of this paper is to identify the main mecanisms that transport pollutants rejected near the coastline by the large city of Marseille and its industrialized suburbs (FosBerre area). The objectives of the paper are adressed using both the dataset collected during the 25 June 2001 ESCOMPTE case (hereafter called J25) and a numerical simulation that was completed with the french non hydrostatic mesoscale model, Meso-NH. This study reveals that three major processes contribute to the transport of matter: the advection of pollutants inland is mainly driven by two breeze cells that occur on two different depth and time scales, and vertical exchanges between the atmopsheric boundary layer and the free troposhere can be generated at the breeze front and near the slopes due to the development of anabatic winds. The horizontal