: In this Thesis, by improvement of multilevel programming model, in the field of kind of relation and effect of upper levels on lower levels, with different look, these problems are solved. This improvement in one state is that uppers level shows partial views for all levels for end condition and in another state each level shows partial views just for his lower level. These problems are modeled with interactive fuzzy goal programming and are solved by using of dynamics programming. In proposed method, the tolerance membership functions in fuzzy sets for measuring the degree of satisfactions of the objectives of levels and degree of achievement of decision vectors controlled by the levels are defined. Then the goal fuzzy programming model for achievement of highest degree of all system, by using dynamics programming is solved. A satisfactory solution is derived when the ratio of satisfactory of levels settled in defined interval. At last, for three, four and five levels problem 50 examples that are random are solved. In two states by using Euclidean distance function, effectiveness of method are comprised. Both states aren’t different. Keywords: Multilevel programming, Fuzzy goal programming, Dynamics programming, Membership function.