We conduct an experiment to test whether probability misperception may be a possible alternative to risk aversion to explain overbidding in independent first-price private-values auctions. The experimental outcomes indicate that subjects underestimate their probability of winning the auction, and indeed overbid. Yet, when provided with feed-back on the precision of their predictions, subjects learn first to predict their probability of winning correctly, and second to curb-down significantly overbidding. The structural estimation of different behavioral models suggests that i) subjects are heterogenous with respect to risk preferences and probability perceptions, ii) subjects tend to best-respond to their stated beliefs, and iii) although necessary to explain fully behavior, risk aversion appears to play a lesser role than previously believed. Finally, our experimental findings are shown to be consistent with a standard theoretical auction model combining risk aversion and misperception of probabilities.