This study shows that horizontal variations in the root zone soil moisture content have an impact on the calculated average surface flux densities and the boundary layer on scales typical of the grid cells of large-scale atmospheric models. This appears even when the land surface is coupled to a developing convective boundary layer. For a typical loam soil type, it appears that in wet conditions, estimates of the latent heat flux density obtained using one uniform root zone soil moisture content are larger than estimates obtained using a distributed approach where the variation in root zone soil moisture content is taken into account. In contrast, in dry conditions using an uniform approach gives lower estimates of the latent heat flux density. As the evapotranspiration is reduced in wet periods and enhanced in dry periods, use of the distributed approach generally gives a weaker seasonal cycle. In this study it is also shown that differences in the estimated surface flux densities lead to differences in the predicted specific humidity in the boundary layer and the predicted atmospheric boundary layer. Thus, taking account of the spatial heterogeneity of the soil moisture content is a prerequisite for a proper representation of the seasonal hydrological cycle within large-scale atmospheric models.