This study is the first to investigate the effect of demand rationing in experimental Bertrand-Edgeworth markets with fixed exogenous capacities. It is found that prices and profits are significantly higher under proportional than under efficient demand rationing. Moreover, the amount of capacity available to each firm is varied. In accordance with earlier studies, prices and profits are significantly higher when capacities are lower. Those effects accord qualitatively with the Nash equilibrium predictions of the corresponding stage games. However, the Nash equilibrium concept does poorly at quantitative predictions. Prices are significantly higher than the Nash prediction in all treatments, irrespective of whether the Nash equilibrium is in mixed or in pure strategies. Profits are higher than the Nash prediction with high capacities, but may converge to the equilibrium prediction in the long run with low capacities. The data of individual price choices feature dynamic patterns that can potentially be explained by both Edgeworth price cycles and imitation of the price set by the competitor.