A reasonable J-resistance curve determination procedure is suggested by correcting the amount of overestimation in J-value whose effect becomes significant when crack grows. Experiments using compact tension specimens of an alloy steel (SCM4, K.S. Designation) are performed. The value of J estimated according to this procedure is compared with that as obtained from the method by Merkle and Corten. The conditions for J controlled crack growth proposed by Hutchinson and Paris are examined for the material used here. It appears that J resistance curve, which is independent of crack ratio can be obtained for those specimens satisfying the conditions of J controlled crack growth.