Central composite design (c.c.d.) robust to three missing observations has been studied by the authors. The performance of a number of central composite designs robust to three missing observations in experimental regions k-spheres and hypercubes at different values of a has been considered. The procedure for the evaluation of loss due to three missing observations has been developed. The purpose of this study is to assess the effects of all possible combinations of three missing observations of factorial, axial and centre points on the size of losses, estimation of the parameters. The minimazloss designs robust to three missing observations called minimaxloss 3 design have been developed for k=2, 3….6.