This paper examines the effect of reputation on auction prices using data from eBay. The authors develop a theoretical model of auctions where bidders and sellers have observable and heterogeneous reputations for default. They find that seller reputation has a statistically and economically significant effect on auction prices, while bidder reputation does not. The study uses data from eBay auctions of Intel Pentium III 500 processors during the fall of 1999. The authors construct an empirical model based on a simple auction model with heterogeneous reputations. They find that bidders pay a statistically and economically significant premium to sellers with better reputations, supporting Shapiro's (1983) theoretical finding of a positive relationship between reputation and price. The authors also examine the effect of bidder reputation on price and find that it is statistically insignificant. They note that previous studies have used data from eBay's feedback system but have not accounted for the fact that feedback profiles are updated in real time. The authors correct for this by using the feedback profiles at the time the auction closed rather than when the data was collected. The authors find that the coefficient of seller reputation is statistically significant, while the coefficients of bidder reputation are not. They also find that the length of the auction has little effect on the final price, but that the variance of the error term is smaller when the auction is longer. The authors conclude that seller reputation is a statistically and economically significant determinant of auction prices. They also note that there are several important issues that this paper has not addressed, including the mechanisms for reputation building and whether reputation is a good predictor of future performance.This paper examines the effect of reputation on auction prices using data from eBay. The authors develop a theoretical model of auctions where bidders and sellers have observable and heterogeneous reputations for default. They find that seller reputation has a statistically and economically significant effect on auction prices, while bidder reputation does not. The study uses data from eBay auctions of Intel Pentium III 500 processors during the fall of 1999. The authors construct an empirical model based on a simple auction model with heterogeneous reputations. They find that bidders pay a statistically and economically significant premium to sellers with better reputations, supporting Shapiro's (1983) theoretical finding of a positive relationship between reputation and price. The authors also examine the effect of bidder reputation on price and find that it is statistically insignificant. They note that previous studies have used data from eBay's feedback system but have not accounted for the fact that feedback profiles are updated in real time. The authors correct for this by using the feedback profiles at the time the auction closed rather than when the data was collected. The authors find that the coefficient of seller reputation is statistically significant, while the coefficients of bidder reputation are not. They also find that the length of the auction has little effect on the final price, but that the variance of the error term is smaller when the auction is longer. The authors conclude that seller reputation is a statistically and economically significant determinant of auction prices. They also note that there are several important issues that this paper has not addressed, including the mechanisms for reputation building and whether reputation is a good predictor of future performance.