This paper uses a unique U.S. airlines panel data set to study empirically the dynamic pricing of inventories with uncertain demand over a finite horizon. I estimate a dynamic pricing equation and a dynamic demand equation that jointly characterize the adjustment process between prices and sales as the flight date nears. I find that the price increases as the inventory decreases, and decreases as there is less time to sell. Consistent with aggregate demand learning and price adjustment, demand shocks have a positive and much larger effect on prices than the positive effect of anticipated sales.