Do prospective customers behave differently based on sellers’ race or signals about sellers’ socioeconomic class? Does this depend on whether a customer lives somewhere racially segregated or plagued by property crime? We investigate these questions in a year- long experiment in which we sold iPods through local online classified advertisements throughout the U.S., each featuring a photograph of the product held by a hand that is dark-skinned (“black”), light-skinned (“white”), or with a wrist tattoo (associated with lower social class). We find that black sellers do worse than white sellers on a variety of metrics: they receive 13% fewer responses and 17% fewer offers, effects similar in magnitude to those associated with the display of a tattoo. Conditional on receiving at least one offer, black sellers receive offers that are lower by 2 to 4%, despite the self-selected—and presumably less biased—pool of buyers. Buyers corresponding with a black seller also behave in ways suggesting they trust the seller less: they are 17% less likely to include their name, 44% less likely to accept delivery by mail, and 56% more likely to express concern about long-distance payments. Black sellers suffer particularly poor outcomes in thin markets; it appears that discrimination may not “survive” in the presence of significant competition among buyers. Furthermore, black sellers do worst in markets that are racially segregated and have high property crime rates, suggesting that at least part of the explanation is statistical discrimination—that is, buyers’ concerns about the time and potential danger involved in the transaction, or that the iPod is stolen goods.