Nine different potential energy surfaces for CO2-Ar taken from the literature are tested for their ability to reproduce the spectra of the Ar-CO2 van der Waals complex and the pressure broadening of CO2 infrared lines by Ar. None of the potentials give a satisfactory account of all the experimental results. All the potentials are found to give significant discrepancies with at least some of the spectroscopic properties of the van der Waals complex. Coupled-states (CS) and infinite-order sudden (IOS) calculations of the pressure broadening cross sections are compared for a few of the potential energy surfaces. The IOS approximation is found to be seriously inaccurate for some potential surfaces, especially for high-j lines, so that CS calculations are essential when comparing with experimental line-shape data. CS calculations of line-broadening cross sections are therefore carried out on all nine different potential energy surfaces. For the pressure broadening coefficients, there are substantial uncertainties in the experimental results. Nevertheless, the only potential to give a satisfactory account of the pressure broadening is the electron gas potential of Preston and Pack, which is the least satisfactory of all for the spectra of the complex. It is concluded that a new potential that reconciles the different data sets is needed.