Pupils in some of Wales's most deprived areas are achieving much better results than wealthier contemporaries, according to new value-added statistics.
In some cases, the figures reverse the traditional ranking of Wales's 22 local education authorities - with deprived Merthyr Tydfil outperforming leafy Monmouthshire at key stage 3 when free school meal entitlement is taken into account.
The results suggest that LEAs in Wales may have more impact on pupil achievement than LEAs in England. Research published there this week suggests they make less than a 1 per cent difference.
Last summer, 54 per of Merthyr's 14-year-olds achieved the standard expected for their age - 8 percentage points more than predicted.
Monmouthshire's pupils topped the raw KS3 tables with 67 per cent, but performed just 1 percentage point above expectations.
Chris Abbott, Merthyr's education director, paid tribute to the hard work of teachers and pupils in producing the best value-added results in Wales at KS3, which he said was an LEA priority.
He added: "The schools ranked us top out of 152 local authorities (surveyed by the Audit Commission) for dissemination of good practice and support for governing bodies.
"If schools are saying LEAs are effective at these things, then these are ways we influence performance in schools."
Alan Pritchard, head of Cyfarthfa secondary school, Merthyr, said it had worked with primary colleagues to track pupil progress from Year 5 and develop a Y5-8 curriculum "so there is no wasted time on entry to secondary school".
Most of its improvement work has occurred internally, but the LEA has helped with well-timed encouragement and funding for important pilot projects, he added.
Neath Port Talbot (NPT) again topped the value-added table for GCSE results, and came second after Merthyr at KS3. In 2005, 57 per cent of its pupils achieved at least five A*-C grade GCSE passes, compared with a predicted 47 per cent.
Kathleen Boyce, head of schools, said schools and officials "continued to pull all the levers" to win positive outcomes for pupils.
She added: "In the smaller authorities the LEA does make a difference, for example, by helping with training and monitoring, and promoting the sharing of good practice and data."
In most authorities in Wales, children are achieving GCSE and KS3 results within 3 percentage points of predictions. But GCSE results are significantly worse than expected in Wrexham and Denbighshire, and at KS3 in Denbighshire. GCSE pass rates are significantly better than expected in NPT, Carmarthenshire, Swansea and Ceredigion.
Professor David Reynolds, of Plymouth university, said such extremes of value-added performance - from 10 percentage points above to 9 percentage points below expectations (Wrexham) - were unlikely in England.
He suggested LEAs in Wales could be more influential because they have more freedom on spending decisions and tend to provide more advisory and support services than those in England. NPT is a relatively high spender on education and, like Merthyr, may also benefit from being small and homogenous. Both LEAs have invested heavily in external school improvement schemes, he added.
Stephen Gorard of York university and Durham's Peter Tymms have concluded that pupil results in England are determined more by wealth and previous attainment than anything LEAs do.
Improvements at individual secondaries are likely to result from changes in intake or the exclusion of difficult students before they sit exams, Professor Gorard said.