Efficiency of power conversion apparatus must usually be measured with very low uncertainty, and depending on the kind of apparatus under test and on the measurement method applied, more or less accurate instrumentation is required. The main issues to be faced, once the target uncertainty is fixed, concern the choice of the measurement method as well as the instruments to be used for fulfilling such requirement. This paper investigates by means of both theoretical analysis and experimental results on the relationship between target uncertainty and instrumentation accuracy in the measurement of uninterruptible power supply efficiency. The results show, amongst other things, that the measurement method allows obtaining very low uncertainty values by means of low accuracy instrumentation.