Computational fluid dynamics (CFD) is an effective technique for investigating atmospheric processes at a local scale. For example, in near-source atmospheric dispersion applications, the effects of meteorology, air-pollutant sources, and buildings can be included. A prerequisite is to establish realistic atmospheric conditions throughout the computational domain. This work investigates the modeling of the atmospheric surface layer under neutral and stable boundary-layer conditions, respectively. Steady-state numerical solutions of the Reynolds averaged Navier–Stokes equations were used, including the k-ε turbulence model. Atmospheric profiles derived from the Cooperative Atmosphere–Surface Exchange Study-99 (Kansas, USA) were used as reference data. The results indicate that the observed profiles of velocity and potential temperature can be reproduced with CFD, while turbulent kinetic energy showed less agreement with the observations. For the stable boundary layer, reasonable agreement of the numerical results with the observations was also obtained for surface layer depth, shear stress, and heat-flux profiles, respectively. The results are discussed in relation to the boundary conditions and sources, and the observation data.