Summary: We prove that two classes of general methods for unconstrained optimization which is constrained by the Fletcher-Reeves (FR) conjugate gradient method are globally convergent under a kind of inexact line search conditions. Results of numerical experiments for several methods are presented.