The loss of efficiency due to one or two missing observations in six factor Box and Behnken design is studied under minimaxloss criterion. The robustness of this design is then compared with six factor central composite designs robust to one or two missing observations.