In exhaust systems with selective catalytic reduction (SCR) a fast conversion of liquid urea to gaseous ammonia and a uniform distribution of the ammonia vapor upstream of the SCR catalyst are essential to reduce the nitric oxides efficiently. For the prediction of the mixing process and the transport of ammonia vapor with the CFD method an accurate description of the turbulent flow field is a basic requirement.This paper presents the comparison of simulation results using three different turbulence models (high-Re kε-RNG model, low-Re kω-SST model, Reynolds stress model) with measurements of the turbulent velocity field using Laser Doppler Anemometry (LDA). The investigations were carried out for a SCR system with a swirl mixer on a cold flow test bench for two different volume flows. From the measured velocity signals different components of the Reynolds-tensor were derived. All three turbulence models are showing a similar mean velocity field as it was expected and there is a good agreement with the LDA data. The turbulence intensity and subsequently the turbulent mixing are underestimated for all simulation setups in comparison to the LDA measurements. However, the anisotropic formulation of the turbulence intensities with the Reynolds stress model (RSM) resolves the flow field more accurately than the two-equation turbulence models.However, further work on a more detailed formulation of the turbulent flow field (DES, LES) is necessary to improve the prediction of the turbulent species transport.