Housing prices have plummeted across the United States. This paper examines differences in the magnitude of housing price decreases across metropolitan areas. A relatively small number of housing market variables observable before the fall are capable of explaining over 70% of the considerable variation in price declines. An additional nonparametric analysis suggests that exceeding particular thresholds for some of the key predictors is associated with much larger price drops. These findings are consistent with historical price patterns and raise questions about the validity of mortgage pricing and risk diversification norms in the US. The analysis points to a set of stylized facts concerning the housing price bubble that need to be explained and suggests fruitful hypotheses for understanding the dramatic housing price declines.