We investigated a chaperone mechanism of protein folding using a 36-mer model on a cubic lattice. The mechanism simulates folding, which proceeds with repetitive cycles of binding, unfolding, and releasing of misfolded metastable states. We measured the yield enhancement due to this mechanism for sequences selected by evolutionary design and showed that the binding and releasing mechanism is efficient for the yield enhancement of folding for sequences that are poorly designed, i.e., where selection is not adequately strong. From this it follows that the chaperone mechanism can be considered as the evolutionary alternative to compensate for poor sequence design. On the other hand, random sequences show a decrease in yield and no effect on the total mean first passage time when the proposed chaperone mechanism is implemented, thus implying that sequence optimization is a necessary condition for the efficiency of the proposed mechanism. We qualitatively reproduced experimental results for folding in the presence of GroEL/GroES, fit our results with the aid of a double-exponential model of folding kinetics, and characterized the conditions under which this mechanism of chaperone action affects folding.