The Guardian has the results. A few words of explanation, since the meaning of the results is far from obvious on a quick glance. The first column gives the number of staff who submitted four pieces of research published 2001-2007 for evaluation. 4* is the highest score awarded, and it's down from there. The number under each score is the percentage of faculty work that received that score. So: the London School of Economics submitted work from approximately 60 faculty (so about 240 pieces of scholarship), of which 45% were deemed to be distinguished at the highest international level (4*). University College London submitted work from not quite 43 faculty (so about 172 pieces of scholarship), of which 35% was deemed to be 4* work. Meanwhile, Oxford submitted work from about 103 faculty (so about 412 pieces of scholarship), of which 35% also received a mark of 4*. That means, of course, that Oxford produced far more 4* work than either LSE or UCL, but its per capita or average performance was not quite as strong overall: that's the "GPA" in the last column, representing the average score for the whole faculty. LSE got a 3.1, UCL got a 3.05, Oxford got a 3.0, Durham a 2.9, and Nottingham a 2.9. (One unknown, to me at least, is whether UCL got credit for the work of two of its most eminent faculty, now both retired: Ronald Dworkin and Basil Markesinis.)
Part of what is going on here is that Oxford has a very large number of staff--the overworked and underpaid tutorial fellows--who have far less opportunity for scholarship. So whereas staff with less onerous teaching loads might produce 8-10 pieces of scholarship during 2001-07, and then have the luxury of submitting the best four for evaluation, it is less likely that many tutorial fellows have that luxury: hence the results.
The big mystery, though, is Cambridge, which includes many of the leading lights of legal scholarship in the U.K. and, indeed, the Anglophone world. Yet Cambridge came in towards the bottom of the top ten based on GPA (2.800), having submitted work from about 83 faculty, only 25% of which received a 4*. My guess is they had a more serious version of the Oxford problem, but it's hard to know. Bear in mind that the scores are awarded by a panel of ten or so legal scholars, who are engaged in (as one colleague in the UK put it) a "speed read" of thousands of pieces of work. All that being said, aside from Cambridge (and also King's College, London, which is weirdly low), the results comport reasonably well with my impressions based on a fair amount of exposure to legal academia in the U.K.
UPDATE: John Gardner (Oxford) analyzes the results in a variety of different ways here.
AND ANOTHER: A legal scholar in Britain writes:
UCL included Dworkin but not Markesinis. The former was still in post at the time of the deadline, whilst the latter was not. (See here http://eprints.ucl.ac.uk/view/rae/38.html )
One of the oddities of the exercise is that an institution will get no credit for someone who leaves the day before the submission date, but full credit for someone who joins the day before. This means there is a perverse incentive to recruit senior staff towards the end of the RAE submission period. Oxford suffered from this for example.
As to Cambridge and King's, I too was initially surprised but less so having reflected upon it. In Oxford and Cambridge the RAE money and prestige is not as relatively important as it is for, say, LSE. There is not therefore the internal pressure to publish there.
The ratings were not just based on submitted work. Such intangibles as research atmosphere and external impact were also taken into account.