Accurate analysis of the homogeneous strain of ellipsoidal objects requires that certain premises are met. These are (1) coaxiality of strain increments, (2) accurate definition of finite strain axes, and (3) perfectly passive behaviour of the objects with respect to the matrix. The equations provide a mathematical model for this idealized scenario.In practice the analysis is limited to data from plane surfaces, and its success depends upon the type of initial fabric of the objects, the frequency distribution of their initial shapes, and errors of measurement of the final shapes (Rf1/2) and of final orientations . Together with the difficulty in locating the finite strain axes, these factors greatly affect the outcome of the analysis, and their combined effects have not been studied previously.Detailed observations on the orientations of strain and of the orientation–distribution of the objects' long axes sometimes indicate that the model is incompatible with the data. Strain analysis is then not possible by the present methods. If the data are compatible with the model and if the original fabric can be shown to have been random or had a preferred orientation parallel to bedding, it is possible to determine the strain accurately to within a few percent. This has been partly confirmed in practical examples in which the strain has been independently determined using the tectonic distortion of rims of individual clasts.