This paper describes an interactive algorithm for fuzzy non linear programming problems. Based on some general results, an iterative algorithm is proposed, which modifies the admissible region in such a way as to increase at each step the global performance. The algorithm can be applied if a crisp sub-problem obtained by the original fuzzy one is solvable. This is the case of quadratic-linear target function with linear constraints. In this case, the numerical simulation showed good convergence property of the proposed procedure.