In this paper the author has considered two-step methods proposed by Ford and Moghrabi with new values of delta. As, it is evident in literature that quasi-Newton methods (single/multi-step) for solution of nonlinear unconstrained problems outperforms all other methods available. In case of multi-step methods it can be noted that different choice of ¿ values effects the performance of algorithm. Furthermore, it can be noted that variables in scalar and gradient space depends upon delta (± ) value. Therefore, in this paper we have considered different choices of ± values to check the performance of algorithm on 15 test problems. Numerical experiments reveals that performance of algorithm improves when ± is taken as 0.85 and 0.95.