This article revisits Paas and Van Merrienboer's (1993) measure of instructional efficiency, which can be applied by educational researchers to compare the effects of different instructional conditions on learning. This measure relied on performance and mental effort on the test, and as such gave an indication of the quality of learning outcomes. The acquisition of more (less) efficient cognitive schemata is indicated by combinations of high (low) performance and low (high) mental effort. This instructional efficiency measure has become widely adopted, but in an adapted form that incorporates mental effort invested in the learning phase instead of the test phase. This article demonstrates that the adaptation has important consequences for the construct of instructional efficiency and for the type of conclusions that can be drawn. Examples are given to illustrate the various implications of different combinations of mental effort and performance measures in the light of more contemporary developments in educational research. (Contains 3 tables and 2 footnotes.)