A procedure to improve the lower and upper bound for the optimal objec
tive function value of a minimization problem is presented. The proced
ure is based on an approximation scheme. Properties of this approximat
ion scheme are established on which the correctness and effectiveness
of the bound improvement procedure are based. The procedure is applied
to improve the computational complexities of approximation algorithms
for several single and parallel machine scheduling problems.