The research excellence framework will be delayed by up to two years if the Conservatives win the 2010 general election, Times Higher Education can reveal.
The Tories intend to postpone it to allow time for a thorough review of plans to measure the wider social and economic impact of research.
The first REF, which will replace the research assessment exercise as the means of distributing £1.5 billion in annual quality-related funding in England, is due to complete in 2013, with up to 25 per cent of departmental scores based on impact.
Speaking exclusively to THE, David Willetts, the Conservative Shadow Universities Secretary, said: "We propose delaying the REF by up to two years to establish whether a sound and widely accepted measure of impact exists.
"If there is a measure that is methodologically robust and widely accepted by the academic community, we would adopt that."
He added that the Conservatives "do not believe that the current proposals pass those two tests. Unless the review is able to establish a measure that does, we would not include impact in the REF."
At present, the Higher Education Funding Council for England, which Mr Willetts said would be invited to carry out the review, is proposing that impact should count for 25 per cent of REF scores, outputs for 60 per cent and the research environment for 15 per cent.
In a consultation on the plans, which closed last month, both the Russell Group of large research-intensive universities and Million+, which represents post-1992 institutions, called for the impact element to be reduced to 15 per cent.
Mr Willetts said it was "possible that impact may be more acceptable to some disciplines than others", also noting the view of some within the sector that it should be measured at the university rather than the departmental level.
He said he was personally in favour of delaying the implementation of the REF because it would allow academics to spend more time on research and teaching.
The Tory plans were warmly received within the sector.
James Ladyman, professor of philosophy at the University of Bristol and the organiser of a petition against the impact proposals, said he was pleased as the current strategy is "deeply flawed", both in terms of the importance given to impact and the plans to reduce the number of panels and units of assessment in comparison with RAE 2008.
"There is a measure for the allocation of research funding that is 'methodologically sound and accepted by the academic community' - namely submitted work as assessed by expert peer reviewers," he said.
The University and College Union, which is not traditionally supportive of the Conservatives, also welcomed the move and called on all the main political parties to "recognise the deep unease within the academic community at these untested proposals".
However, Bahram Bekhradnia, director of the Higher Education Policy Institute, said Mr Willetts had been "sensible" not to commit to abandoning impact altogether.
He added that Hefce had been "steamrollered" into committing to measuring and rewarding impact before it had established how this could be done.
"What is needed is a proper review of how - and whether - impact can be measured, the value to be given to it in the REF and the balance between academic and non-academic impact," he said.
Geoffrey Crossick, warden of Goldsmiths, University of London, was more positive about the current situation, but nevertheless welcomed the postponement.
Hefce was developing a "case-study approach" that would "capture the breadth of the impacts in the arts, humanities and social sciences", he said.
"I would be worried about any single measure for impact, however methodologically sound, so a new approach would have to be responsive to that breadth," he said.
"A postponement would certainly make it easier for Hefce to get the new approach right and for universities to prepare."
As THE has reported, alongside the concerns about the inclusion of impact in the REF, there has been a groundswell of opposition to the new requirement that scholars detail the potential impact of their proposals in research council grant applications.