Context. Radiative transfer calculations have predicted intensity enhancements for optically thick emission lines, as opposed to the normal intensity reductions, for astrophysical plasmas under certain conditions. In particular, the results are predicted to be dependent both on the geometry of the emitting plasma and the orientation of the observer. Hence in principle the detection of intensity enhancement may provide a way of determining the geometry of an unresolved astronomical source. Aims. To investigate such enhancements we have analysed a sample of active late-type stars observed in the far ultraviolet spectral region. Methods. Emission lines of O vi in the FUSE satellite spectra of Eri, II Peg and Prox Cen were searched for intensity enhancements due to opacity. Results. We have found strong evidence for line intensity enhancements due to opacity during active or flare-like activity for all three stars. The O vi 1032/1038 line intensity ratios, predicted to have a value of 2.0 in the optically thin case, are found to be up to ∼30% larger during several orbital phases. Conclusions. Our measurements, combined with radiative transfer models, allow us to constrain both the geometry of the O vi emitting regions in our stellar sources and the orientation of the observer. A spherical emitting plasma can be ruled out, as this would lead to no intensity enhancement. In addition, the theory tells us that the line-of-sight to the plasma must be close to perpendicular to its surface, as observations at small angles to the surface lead to either no intensity enhancement or the usual line intensity decrease over the optically thin value. For the future, we outline a laboratory experiment, that could be undertaken with current facilities, which would provide an unequivocal test of predictions of line intensity enhancement due to opacity, in particular the dependence on plasma geometry.