Using previously described close coupling (CC) and coupled states (CS) cross sections for N-2-N-2 collisions [J. Chem. Phys. 104, 7572 (1996)], we have calculated CARS linewidths at room temperature and below, Agreement with experimental values at room temperature is quite good but predictions become increasingly too large at lower temperatures, with errors reaching 10%-30% at 113 K. We believe these low temperature discrepancies reflect errors in the intermolecular potential used here. To obtain line widths at higher temperatures we have used the energy corrected sudden (ECS) approximation, taking the fundamental cross sections, sigma(0.0-->J(1), J(2)), from the CC-CS calculations extended to higher collision energies with additional CS and infinite order sudden (IOS) calculations : the ECS scaling distance, I-c was chosen by fitting to the 300 K CC-CS results. In general, we find rather good agreement with experimental values to 1500 K, although it appears that smaller values of I-c are more appropriate for higher temperatures and for higher rotational levels. This variability of I-c is reasonable from physical arguments but somewhat diminishes the predictive utility of this approach. Agreement of these purely ab initio predictions with experimental data is nearly as good as that obtained from the best rate law model whose parameters were fitted to these data.