Hydrodynamics in an oxidation ditch is a main key of an optimal treatment. Mixing and aeration are closely linked to biological performances. One parameter considered to be essential for correct design and operation is the horizontal velocity without aeration. A value of 0.3-0.35 m.s -1 for this parameter is regarded as a prerequisite (Da Silva, 1994). Hence its correct prediction is high importance. Different CFD modelling approaches have been applied and evaluated with regard to their performance in modelling the agitation in an oxidation ditch and consequently assessing the horizontal velocity. The two approaches used differ in the way they take the agitation, i.e. the impeller, into account in the model set-up : a) the "Fixed Values" approach, imposing a local axial velocity on cylinders which represent the impellers; b) the "Multiple Reference Frame" approach (MRF), using the exact geometry of the impellers. Theses approaches are compared with experimental data, i.e. with the average horizontal velocity as well as with velocity profiles. The first CFD agitation model ("Fixed Values") is highly dependant on the quality of the information provided by the supplier of the impellers. This is due to the fact that the supplier presents the value of the imposed axial velocity for a given impeller. Hence this whole modelling approach depends on the accuracy of the value provided. However, once applying the correct axial velocity imposed, this approach can gain quite some relevance as it is less computational intensive than the MRF approach . Experimental measurements have in fact revealed a too low horizontal velocity in the oxidation ditch of concern. Thanks to the CFD simulations, presented in this study, explanations of the low horizontal velocity could be put forward. This illustrates their potential for advanced troubleshooting.