Abstract Cognitive load theorists have frequently used subjective measures of cognitive load to test the effectiveness of instructional procedures. This study sought to broaden the applications of subjective measures by testing their ability to detect variations in intrinsic cognitive load within tasks. In two experiments students were asked to complete algebraic problems and provide a subjective measure of cognitive load for each computation completed. By keeping extraneous and germane cognitive load constant, it was argued that changes in intrinsic cognitive load (element interactivity) were measured. Results showed that subjective measures were highly reliable, varied significantly within problems and correlated highly with errors. Evidence was also found that the subjective measures were influenced by the expertise of the learner and procedural errors.