Academics have challenged ministers' claims of improved pupil performance in national tests and GCSEs.
Government research obtained by The TES compares the results of pupils in England with teenagers in other countries. The findings weaken ministers'
claims that pupils are getting better at English, maths and science.
The analysis came as the proportion of top-grade A-levels rose this week by its second-highest amount since the A grade was introduced in 1965. Key stage 2 results, to be published next week, are also expected to rise.
Ministers have long hailed the years of rising test and exam results as a success story for Labour, rejecting suggestions of dumbing down.
The analysis was carried out by academics at Southampton university. It looked at the performance of England's students in the best-known tests used to rank the quality of developed nations' education systems.
The Organisation for Economic Co-operation and Development's programme for international student assessment (Pisa) tests are taken by around 250,000 15-year-olds in 41 countries.
The analysis found evidence that pupils who had achieved average results in key stage 3 tests in English, maths and science and GCSEs performed worse in the 2003 tests than those in 2000.
The average for students in all Pisa countries in the tests is 500 points.
In 2000, English students with average KS3 results achieved 523 points in reading. Their counterparts in 2003 scored only 506.
There were similar falls in maths and science. Professor John Micklewright, who led the study, said one possible explanation was that the standard of the KS3 tests and GCSEs had declined, though there was no proof.
But Professor Alan Smithers, of Buckingham university, said: "This study calls into question the Government's claims that there have been continuous improvements in our education system under its tutelage."
And Professor Peter Tymms, of Durham university, said: "This raises questions about how genuine the test improvements are."
The Government has rejected the 2003 results, which were given an unofficial status because too few English schools agreed to take the Pisa tests. It said that this could have biased the findings.
But the Southampton report said the bias was in England's favour. In both 2000 and 2003, the pupils who took the Pisa tests had above-average KS3 and GCSE results. If the results had been adjusted to take this into account, England's position in 2000 would have slid from eighth to ninth in reading; and from fourth to seventh in science. Maths would have been unchanged at eighth.
In 2003, results would have slid by one place in each of reading, maths and science, from unofficial results of 12th, 18th and 11th.
The Department for Education and Skills said: "Pisa and key stage tests measure different things. Performance against the national curriculum through the national tests have shown significant improvements."
It said an alternative international study of secondary pupil performance in maths and science had shown there had been no decline for England between 1995 and 2003.