The concept of «quality», applied to research evaluation, is being criticised first as it is conceived as absolute, instead of highly specific to public expectancies; secondly, because – contrary to the scientometric approach – quality can be evaluated, and not measured; thirdly, the definition of research quality is theory-laden, or better, context-laden. The article stresses 6 types of context, relevant to shape the concept and evaluation of research quality: 1. the type and level of research finalization; 2. the level of aggregation of social expectancies; 3. the systemic and institutional context of research; 4. the socio-cognitive subsystem (or, in other words, the disciplinary matrix) of research; 5. the specific relevance of research contributions in reference to scientific expectancies; 6. the kind of decision to be taken as a consequence of the evaluation process. The A. concludes rejecting the scientometric approach to research evaluation, inasmuch scientometric indices measure the productivity and/or the diffusion, but not the «quality», of research, while deferential citing, overpublishing, self-citation, and lobbyist citation can alter the functioning of the recognition of excellence system.