The EUROMET project 428 examines efficiency transfer results for Ge gamma-ray spectrometers when the efficiency is known for a reference point source geometry. For this, different methods are used, such as Monte Carlo simulation or semi-empirical computation. The exercise compares the application of these methods to the same selected experimental cases to determine the usage limitations versus the requested accuracy. For carefully examining these results and trying to derive information for improving the computation codes, this study was limited to a few simple cases. The first part concerns the simplest case of geometry transfer, i.e., using point sources for 3 source-to-detector distances: 2, 5 and 20 cm; the second part deals with transfer from point source geometry to cylindrical geometry with three different matrices. The general conclusion is that the deviations between the computed results and the measured efficiencies are mostly within 10%. The quality of the results is rather inhomogeneous and shows that these codes cannot be used directly for metrological purposes. However, most of them are operational for routine measurements when efficiency uncertainties of 5-10% can be sufficient.