With the methods ordinarily employed of establishing the efficiency of the turbo-alternator, the stray loss that is present on load is either deduced by a substantially conventional method or is ignored altogether. As alternators for larger outputs are constructed, the stray loss becomes of greater importance and, indeed, may prove to be one of the factors that will limit the maximum size of unit that it is desirable to employ. A still more unsatisfactory feature of the existing methods is that it is becoming impracticable to apply them to the very large alternators now being constructed, on account of the difficulty of providing a driving motor of sufficient power. Further, with a driving motor, it is obviously impracticable to make any tests on site. It becomes increasingly important, therefore, to introduce a method of establishing the alternator efficiency which is based on the measurement of the actual losses on load.The object of this paper is to show that the actual losses on load can be deduced conveniently and accurately from measurements of the cooling air flowing through the alternator. It is further shown, however, that although such methods are sound in principle, misleading results may be obtained unless certain precautions are taken in applying them.