The MIND (Method for analysis of INDustrial energy system) method has been developed for multiperiod cost optimization of industrial energy systems. Existing industrial processes can be represented at the desired level of accuracy, i.e. one modeling unit may represent a part of the production process or the whole plant. The optimization method includes both energy and material flows. Nonlinear relations, energy conversion efficiencies and investment costs are linearized by mixed-integer linear programming. A flexible time-scale facilitates the performance of long- and short-term analyses. In order to meet the requirements with regard to sustainable development, the recycling of energy and material flows is becoming more common in many industrial processes. The recycling or reuse of energy and material is managed by feedback loops, which are incorporated into the original MIND method to improve the model and reduce the calculation time. The improved MIND/F method (MIND method with feedback loops) model is applied to a pulp and paper mill in Sweden. A comparison between the original MIND method with manual handling of the feedback loops and the MIND/F method gives highly satisfactory results. Cost optimization using the improved MIND method is well within the given accuracy and computer time and manual calculation time are both reduced considerably. The reuse of energy and material resources is not only an economic advantage, but also implies a reduction of the environmental impact. (C) 2003 Elsevier Ltd. All rights reserved.