Annual, aggregated GCSE scores for councils - published in league tables every autumn - do not count the results of those at pupil referral units.

Few of these pupils achieve good grades; the Government’s policy of leaving them out of the calculations is therefore marginally improving authority scores.

A Department for Education and Employment spokesman said that any benefit to education authorities was “statistically insignificant” because, nationally, only 1,700 pupils out of 581,000 taking GCSEs each year were affected.

But the admission was seized upon by one of the country’s largest authoriies, which argued that excluding PRU pupils from the calculation was out of step with ministers’ social inclusion policies.

Last year, Birmingham calculated its own GCSE score, which included the results of all pupils at PRUs. It found that 37 per cent of its pupils gained five or more GCSEs, one percentage point lower than the Government’s figure.

Dr John Hill, Birmingham’s research and statistics manager, said the council was happy to accept the lower figure, which had been circulated to the local media.

He said: “We as an LEA want to make sure that all of our students get some qualifications. Therefore, when the Government and the public assesses an authority’s performance, we think they should include all pupils.”