This paper estimates Sutton's lower bound [1991, 1998] by quantile regression, and thus shows the influence of outliers on previous estimates that used the simplex method. The lower bound estimates are obtained separately for exogenous and endogenous sunk cost industries in Italian manufacturing sectors in 1995, using microdata from the SCI’95 (Firms Accounts System) survey conducted by ISTAT (National Institute of Statistics, Italy). The results suggest that Sutton's predictions are robust.