The purpose of this paper is to compare sampling performance of the minimax generalized ridge regression estimators considered by Casella (1985) with that of ordinary least squares estimator by numerical calculations of exact mean squared error of these estimators.