The development of systems based on Stirling machines is limited by the lack of data about the performance of the various subsystems that are located between the input and output power sections. The measurement of some of the variables used to characterise these internal subsystems presents difficulties, particularly in the working gas circuit and the drive mechanism, which causes experimental reports to rarely be comprehensive enough for analysing the whole performance of the machine. In this article, we review experimental data from a V160 engine developed for cogeneration to evaluate the general validity; we also investigate one of the most successful prototypes used in dish-Stirling systems, the V161 engine, for which a seemingly small mechanical efficiency value has been recently predicted. The procedure described in this article allows the possible margin of improvement to be evaluated for each subsystem. The procedure is based on similarity models, which have been previously developed through experimental data from very different prototypes. Thermodynamic models for the gas circuit are also considered. Deduced characteristic curves show that both prototypes have an advanced degree of development as evidenced by relatively high efficiencies for each subsystem. The analyses are examples that demonstrate the qualities of dimensionless numbers in representing physical phenomena with maximum generality and physical meaning. (C) 2014 Elsevier Ltd. All rights reserved.