The MHD stability in a tokamak, especially the Mercier criterion, is quite sensitive to the equilibrium parameters. In the paper, we solved the free-boundary equilibrium with hollow current profile and, in usual cases, the code gives the solution with enough accuracy. Through a careful re-examination of the equilibrium, however, it was found that the equilibrium accuracy was reduced for increasing q-profile hollowness and, near the magnetic axis, the plasma occasionally tended to the inverse D shape, which reduces the Mercier stability index. This resulted in the misleading output of the paper; that is, the Mercier criterion can be violated for the high inverse q profile even for q0 greater than 1. By the re-calculation of the Mercier criterion for equilibria with high accuracy, we conclude that, although the stability index reduces with increasing q0, the Mercier criterion is not violated. We acknowledge many colleagues who pointed out the importance of the equilibrium accuracy and suggested the re-examination of the stability calculation.