The effectiveness of five different turbulence models is assessed for the flow across a row of confined cylinders at a pitch-to-diameter ratio of 1.7 and at Reynolds numbers ranging from 2621 to 55 920. Models examined include the one-equation Spalart-Almaras model; two-equation realizable k - [curly epsilon], k - , and shear stress transport models; and a four-equation v2 - f model. Quantities compared against published experimental data include minor loss coefficients, separation angles about cylinders, wake lengths behind cylinders, and streamwise velocity profiles at the periodic inlet/outlet boundaries. Results indicate that each of the models did a reasonable job in predicting the minor loss coefficient as a function of Reynolds number. With the exception of the k - [curly epsilon] model, each was also able to predict the experimentally observed trend of decreasing wake and separation lengths with increasing Reynolds number. In addition, all models also predicted a local minimum in the separation angle about the inner cylinder as a function of Reynolds number, which has also been observed experimentally. Our conclusion is that the v2 - f model performed slightly better at predicting the experimental data than any of the other models examined, although at the computational expense of solving two additional equations.