Self-report is required to assess mental states in nuanced ways. By implication, self-report is indispensable to capture the psychological processes driving human learning, such as learners' emotions, motivation, strategy use, and metacognition. As shown in the contributions to this special issue, self-report related to learning shows convergent and predictive validity, and there are ways to further strengthen its power. However, self-report is limited to assess conscious contents, lacks temporal resolution, and is subject to response sets and memory biases. As such, it needs to be complemented by alternative measures. Future research on self-report should consider not only closed-response quantitative measures but also alternative self-report methodologies, make use of within-person analysis, and investigate the impact of respondents' emotions on processes and outcomes of self-report assessments.