Three different optimization procedures, which are expected to improve the performance of large solar thermal systems, are described in this paper. The first procedure concerns the planning phase of the system. The second one considers its operation and should be carried out after about one year of data collecting. The third one examines the daily performance considering predictions of weather and hot water consumption and the actual temperature level in the buffer store. For all three optimization steps. the feasibility and the energetic potential have been investigated. For these studies validated system models of two solar domestic hot water systems in Germany have been implemented in TRNSYS. In combination with the simulations, both classical algorithms and Evolutionary Algorithms have been applied, for the optimization procedures. (C) 2003 Elsevier Science Ltd. All rights reserved.