Modern condition of management theory and practice of operating costs of railway enterprises may be characterised like transition, when traditional methodologies demand corrections, taking into account changes in connection with realization of the reform processes and structural conversions. Traditional management theory and the system of every type ( economic, technical) are based on the principle of feedback. Managerial decision is based on analysis of system respond to internal and external influences. Generally, the subject of management in the economic system is a manager of a specific level, '' armed'' with the knowledge base about the conceivable behaviour of the system depending on internal and external characteristics. The inertial behaviour of the economic system ( enterprises) in conditions of stable economy allows using their behaviour trend models, correlative and regression analysis for prognostication. Unfortunately, in the recession situation or structural changes in the control object ( amalgamation, separation), the retrospective foundation of information and knowledge base of the manager loses its topicality for using traditional prognostication and administration methods. Let us assume that the original object management ( company) O is divided into two objects ( companies) O1 and Omega 2, Omega= Omega 1. Omega 2 and Omega 1. Omega 2= circle divide. We assume that there is no synergistic effect of association. Then the set of input parameters x will also be divided into two subsets x= x1 boolean OR x2. In this case, the retrospective basis to manage the new enterprise Omega 1 will be x1, and the second will be x2. It is natural to presuppose that the further dynamic development of the first enterprise depends on retrospective database indicators x1. Herewith, the control system is described as x1= F( X, Y), x2 = U ( X, X1, Y), where F is a functional determining the results of the enterprise under the influence of the internal input and external factors X and Y; U is a functional determining the control impact on the analysis of output data of the enterprise. Let X = ( x1, x2,., xn) be the vector characterizing the absolute values of the parameters A = ( a1, a2,., an) of the system -Omega. Division of the enterprise into two structures Omega 1. Omega 2 suggests new subsets A1 and A2, in which each of the indicators ai: 1) a(i) subset of A(1), a(i)is not an element of A(2); 2) a(i) is an element of A(2), a(i) is not an element of A(1); 3) ai is an element of A(1), ai is an element of A(2). In general, we can assume that xi1 = alpha 1ixi; xi2 = alpha 2ixi; alpha 1i + alpha 2i = 1, where xi1 is the share of index xi, attributed to the structure Omega 2, xi2 is the share of index xi, attributed to the structure Omega 1. The previously proposed concept of operating costs prediction and control based on the mechanism of correlative and regression analysis. Thus, regression function can be transformed to control operating costs of rail transport company in terms of structural changes (separation).