The energy performance in buildings is a complex function of the building form and structure, heating system, occupancy pattern, operating schedules, and external climatic conditions. Computer simulations can help understand the dynamic interactions of these parameters. However, to carry out a multi-parameter analysis for the optimisation of the building energy performance, it is necessary to reduce the large number of tests resulting from all possible parameter combinations. In this paper, the lattice method for global optimisation (LMGO) for reducing the number of tests was used. A multi-parameter study was performed to investigate the heating energy use in office buildings using the thermal simulation code APACHE (IES-FACET). From the results of the sensitivity analysis it was possible to estimate the relative importance of various energy saving features. (C) 2001 Elsevier Science B.V. All rights reserved.