The policy, which RCUK hopes to adopt by the summer, stipulates that the final version of papers produced with funding from any of the science research councils must be made freely available online within six months of publication.
Research funded by the Economic and Social Research Council and the Arts and Humanities Research Council would have to become open access within 12 months. RCUK would hope to see this period reduced to six once publishers in these fields, which are often smaller than science publishers, were ready to make the transition.
But the Publishers Association, which represents UK publishing companies, criticised the proposals and said it had not been consulted.
It said in a statement that "more in sadness than anger, we have little option but to oppose this policy".
"No evidence or impact assessment is offered for the effect of six-month embargoes on the large majority of articles published [in] subscription [journals].
"[The policy] takes no account of the role of publishers in scholarly communication, makes no reference to sustainability or the management of peer review, offers no practical policy for funding [author-pays] open access while dictating firm and onerous requirements for mandatory deposit on short embargoes."
The association also questioned the timing of the circulation of the draft, saying it was "clearly" designed "to undermine the collaborative work of the Finch group on extending access".
That group, chaired by former Keele University vice-chancellor Dame Janet Finch and convened by the government last September, was charged with bringing publishers, funding bodies and universities together to find a way to implement open access. It is expected to report in May.
A spokeswoman for RCUK said that members of the group, among others, would be consulted.
"The research councils will seriously take on board any feedback and the recommendations from the Finch group before deciding on any final policy," she said.
The proposed six-month embargo period is shorter that the one-year period imposed by the National Institutes of Health in the US.
Open-access mandates would have been outlawed by the recently abandoned US Research Works Act. That bill was initially supported by Elsevier, but the publishing giant withdrew its support last month in the wake of a pledge to boycott the firm by more than 8,500 academics.
RCUK has had an open-access policy since 2006, but the Medical Research Council's six-month requirement was the only stipulated maximum embargo period. The Wellcome Trust, a biomedical funding body, also insists on a maximum embargo of six months.
Peter Murray-Rust, a reader in the Unilever Centre for Molecular Science Informatics at the University of Cambridge and an open-access advocate, described RCUK's policy as a "major step forward" that would provide a template for other funders and mark "the beginning of the end" for restricted access.
"It is going far further than the publishers will accept, but [by not insisting on immediate open access] RCUK is also showing it isn't behaving unreasonably," he said.
He particularly welcomed the requirement for publishers to permit data-mining and the unrestricted reuse of content.
But another prominent advocate of open access, Stevan Harnad, professor of psychology at the University of Quebec Montreal, described these two stipulations as "premature" and "over-reaching" in the absence of immediate open access.
"It is like a starving nation discussing whether it would like to have steak Milanese when it is not even getting basic grub," he said.
• The draft RCUK policy is available here.