The question of whether individual judgment errors survive in market equilibrium is an issue that naturally lends itself to experimental analysis. Here, the Monty Hall problem is used to detect probability judgment errors both in a cohort of individuals and in a market setting. When all subjects in a cohort made probability judgment errors, market prices also ref lected the error. However, competition among two biasfree subjects was sufficient to drive prices to error-free levels. Thus, heterogeneity in behavior can be an important factor in asset pricing, and further, it may take few bias-free traders to make asset prices bias-free.IN RECENT YEARS, behavioral researchers have challenged the joint hypotheses of efficient markets and the idea that average (excess) returns are solely rewards for risk, suggesting that these are deficient for interpreting capital market data (De Bondt and Thaler (1985), Shleifer (2000), Hirshleifer (2001)). They argue that individual cognitive errors are useful in interpreting marketwide anomalies, presupposing that individual judgment errors aggregate into similar marketwide errors. Other researchers instead maintain that individual judgment errors do not materially affect markets in equilibrium. Rubinstein (2001, p. 26) summarizes: "Some adherents of behavioral finance begin sensibly enough with the results of convincing experiments that show human beings are irrational in certain specific systemic ways. But then comes the hand waving as they try to extend the results to the much more complex, long-lasting, repetitive and subtle environment of the market. This extension requires a big leap of faith." This paper offers direct experimental evidence about the link between individuals' judgment errors and market equilibrium. We perform two sets of experiments: the first establishes the absence or presence of judgment errors in a cohort of individuals. In the second experiment we change the context to a market setting. We can then observe the extent to which judgment errors observed in the individual experiments are ref lected in market prices and allocations. We further vary the types of trading opportunities to test some conjectures about how judgment errors could be eliminated in markets.