The ESRC Teaching and Learning Research Programme is expected to produce knowledge of high social scientific quality and practical pedagogic value. It faces, in a particularly acute form, the issue of how confidence in its conclusions can be built. This article begins with some conceptual clarification of the notion of a 'warrant' for research findings and distinguishes between warrants in social research and warrants for educational policy and practice. It then, in the specific context of research on learning outcomes, proposes an integrated model that recognizes the importance of, and relationships among, criteria associated with research design and empirical work, theoretical coherence, relationship with other research, and user affirmation. It concludes by emphasizing the need for continuing debate about, and development of, the roles and relationships of 'researchers' and 'users'.