The Wellcome Trust, the UK's biggest research charity, this week came out in favour of a wholly quantitative metrics-based research assessment system to replace the current exercise.
Mark Walport, Wellcome's director, said a metrics system would suit the biomedical sciences, with which it is heavily involved. He said that the peer review involved in vetting research grants would be enough to ensure the quality of research.
"The most peer-reviewed activities are grants from charities and research councils. It's also competitive and international. For the biomedical sciences, grant income is a relevant metric," he said.
But others disagree. Alex Markham, chief executive of Cancer Research UK, said: "The published output of research is extremely important. There are very able researchers who produce fabulous output with very little money, and people who produce very little research with extraordinarily large amounts of money. Ignoring the output metric is fraught with danger."
Lee-Ann Coleman, head of research policy at the Association of Medical Research Charities, said: "All our members are committed to peer review. Inputs do not necessarily mean outputs."
Learned societies are also determined to keep peer review.
Maths societies said metrics alone would not produce reliable judgments.
They called for a more frequent exercise to avoid "disruptive" gaps between assessments.
The British Academy said research assessment needed to be slimmed down but a hasty move to metrics would be "ill-advised". A wholly metrics-based approach, especially one based on external grant income, could discourage long-term research in the humanities and social sciences.
Malcolm Cook, chair of the Modern Humanities Research Association, said:
"The humanities need something a lot more nuanced."
The Academy of Social Sciences called for "a carefully selected combination of quantitative and qualitative methods". It said the difficulty lay in identifying quantitative measures that reflected quality.