Optimisation models are commonly used in analysing and selecting the optimal configuration of a cogeneration or trigeneration system. Due to the high variability of energy demand in the residential and tertiary sector, hourly demand data is needed in order to accurately analyse and optimise energy supply systems, but a very high number of hours leads to very high or unfeasible computational time expenses. To overcome this problem, the most common practice is to reduce the number of days used in the optimisation model working only with typical days, but there are no guidelines on how to select these days and it is not clear if these typical days properly represent the whole year in the optimisation model. This study proposes a method for the selection of typical days of hourly energy demand for a whole year and analyses the influence of the results on an optimisation model for a trigeneration system. Heating and cooling energy demand data has been obtained from energy simulation software for buildings and an optimisation model for a trigeneration system has been developed. The optimisation of the whole year is then compared with the optimisation of the selected typical days. The results using typical days are very similar to the results of the whole year. (C) 2010 Elsevier Ltd. All rights reserved.