This paper investigates the input-output and the calorimetric methods of evaluating stray load loss in induction motors. It highlights the difficulties of accurately measuring stray load loss experimentally and thereby declaring the correct machine efficiency. The standard experimental approach of input-output power measurement with loss segregation is widely used but has some limitations. Fifteen induction motors ranging between 11 kW (15 hp) and 200 hp have been carefully studied in this paper mainly using IEEE 112 Method B. Additional results obtained by calorimetric methods have been used to validate the standard test methods. The objectives of this paper are to quantify the stray load loss, to identify possible sources of error in such a process, and to provide indications of how some of these errors may be mitigated. The conclusion drawn confirms that the only valid way is to evaluate stray load loss directly and any arbitrary allowance for stray load loss is unfounded. The importance of this work is in improving measurement techniques and instrumentation accuracy.