The management of a manufacturing system faces with the problem of long-term and medium-term forecasting of the system-forming factors of the stable functioning of the system(raw material prices, expected demand, the volume of orders, cost of products), changing due to dynamics of the markets for raw materials and end-products, high competition, as well as other political, economic and global factors. The expected improvement of the forecasting quality has not been achieved even with a large amount of qualitative and quantitative information about the system. The events accompanied by cumulative influences from many factors often lead to rapid changes in prices (structural shifts, changes in trends and price relationships) that are reflected in the quantitative data with a delay. Analysis of the strength and direction of cumulative influences of many factors on prices improves the quality of forecasting. This paper proposes the general scheme of the forecasting algorithm based on a system of models that includes a situation cognitive map and a set of the quantitative forecasting models of changes in the key indicators of a manufacturing system