When buildings are subject for refurbishment, it is very important to add the optimal strategy at that very moment. If other solutions are chosen and implemented, it will no longer be possible to change the building at a later occasion with the same profitability. A suitable criterion for optimality is the point where the life-cycle cost (LCC) has its minimum value. This point can be calculated by using so-called mixed integer linear programming (MILP). This paper shows how building and possible fenestration retrofits are described in such a MILP program. Changing existing double-glazed windows to triple ditto will of course make the U-values lower, but at the same time less solar radiation is transferred through the glass panes. This must be properly addressed in the MILP model. Of vital importance are also the heating system and the energy tariff connected to it. Nowadays, time-of-use rates are common practice both for district heating and electricity. These facts make it unsuitable to write, optimise and solve the MILP model "by hand", and instead a computer program has been designed for writing the model in the form of a standard MPS data file. This file can in turn be scanned and optimised by MILP-solving programs available at the market today. © 2001 Elsevier Science B.V. All rights reserved.