Most people in mountainous areas reside in valleys where air pollution frequently poses serious problems. Our understanding of vertical exchange processes in the surface and boundary layers of valleys and their effect on the dispersion of air pollutants is limited. Characteristics of the exchange processes are largely determined by turbulent surface heat fluxes and the convective boundary layer structure. The MAP-Riviera field campaign produced a unique meteorological data set in the Riviera Valley for studying the valley boundary layer. This 3-km wide valley in the Swiss Alps is characterized by heterogeneous landuse and local slope angles exceeding 30 degrees. The data set includes turbulence data obtained at nine different measurement sites on the valley floor and sidewalls, as well as aircraft and rawinsonde data. In addition to data-analyses, the Regional Atmospheric Modeling System (RAMS) is used to determine exchange processes in the valley atmosphere. This study focuses on the special observation period of 25 August 1999, a fair weather day on which thermally driven wind systems were well-developed. The three-dimensional simulations use two-way interactive, nested grids, with the innermost grid having a grid size of 333m. The simulations are initialized with gridded pressure data from ECMWF model analyses and data from the European rawinsonde network. High resolution land-use and soil-type data, and soil moisture output from a sophisticated hydrological model are used at the lower model boundary. Interesting features of the daytime boundary layer structure include a spatially inhomogeneous wind field with a wind maximum occurring on the eastern side of this north-south running valley. The mesoscale model proved to be very useful in explaining this phenomenon and indicates that interactions between terrain-forced and thermally driven flows played an important role. Modeled surface heat fluxes and thermodynamic structure of the boundary layer generally agree well with observations but there are important differences that will be discussed in detail. Difficulties associated with the evaluation of modeled turbulent surface heat fluxes and boundary layer structure in highly complex terrain will be addressed. Implications of the modeled and observed boundary layer structure for the dispersion of pollutants in valley atmospheres will also be discussed.