This paper investigates the effectiveness of airport slot auction by developing a two-stage model, which distributes slots among competing airlines using an ascending-bid multi-unit auction. Airlines assign different values to slots at peak and off-peak periods, and each carrier has its private (subjective) forecast of future demand. Conditional on the slots available to them, airlines compete in frequency, fare, and aircraft size over a congested airport network. The market outcome under such a scheme is benchmarked to those under an ex ante allocation and an ex post allocation by a social planner. Comparison results suggest that the auction-based scheme is inferior to the other two schemes when there is little fluctuation in demand, whereas the auction-based system is more effective when there is substantial demand uncertainty. Auctioning some grandfathered slots can improve social welfare but the marginal effect may diminish quickly.