We analyze how the seller adjusts the reserve price in infinitely repeated auctions using the information conveyed by past bids. Bidders are myopic and have constant valuations; losers are replaced by new bidders, and winners leave with an exogenous probability. Our model is a stylized description of the market for online display advertisements, where publishers sell impressions through real-time first- or second-price auctions. The optimal reserve price is either equal to the value of the last winner, or lower than it when the winner’s value is sufficiently high. In this second case, the reserve price decreases in the winner’s value in a first-price auction, while it is independent of it in a second-price auction. Because past winners who are outbid substitute for the reserve price in a second-price auction, the seller often sets a lower reserve price and obtains a higher revenue than in a first-price auction. Long-run trade may be non-monotonic in the probability that winners leave.