This paper presents a new penalty functions method for nonlinear programming with the change of penalty parameter at each iteration. The efficiency of the method is illustrated by a number of test problems.