It shows that many of the 2,000 external markers were not adequately qualified or trained; that they did not always have enough knowledge of teaching the age group, and that they did not have enough time to mark properly.
The evaluation raises questions about how enough good markers can be recruited and it will lead to renewed calls for teachers to mark their own pupils' tests.
The report also shows that markers were reluctant to give grades at the top of the scale or to give full marks, even on work which fulfilled all the criteria.
This finding supports complaints by teachers. Some 13 per cent of schools appealed over English results and 4 per cent had levels regraded.
The School Curriculum and Assessment Authority has already admitted there were problems and agreed to tighten up the criteria and training.
Statistics published last month showed 55 per cent of 14-year-olds achieving level 5 and above on the English test and 28 per cent getting level 6 and above. The "typical" 14-year-old should be at one of these two levels, according to the Government.
Although SCAA figures show that 94 per cent of the markers had recent or current English teaching experience, the in-depth study of the work of 17 randomly-chosen markers by Exeter University researchers shows that only 9 - just over half - were currently secondary English teachers.
The others consisted of four supply teachers, one primary teacher, two secondary teachers of other subjects and a retired RE teacher.
The report, examining the external marking of English at 14, says that "within this limited sample, those who do not teach English as subject specialists displayed inaccuracies andor inconsistencies in their marking".
But it also says that the best markers were those with the least examining experience of GCSE English. GCSE markers are accustomed to a higher level of work and a different type of test.
The best markers were those teaching English to the age group being tested, and the report recommends that these are the people who should be appointed for next year's tests.
However, Anne Barnes, general secretary of the National Association for the Teaching of English, is not convinced they will be persuaded to take on the job.
"Teachers of key stage 3 do not want to be marking 250 papers of other schools' work," she said.
A linked report on key stage 3 assessment by the Exeter team, led by Dr Hilary Radnor, says the controversial literature paper on Shakespeare should be replaced by coursework marked by the pupils' class teachers, with external moderation.
Dr Radnor told The TES: "Taking assessment out of the control of teachers is against high-quality learning."
A great deal of money had been spent developing good assessment instruments, and if teachers were more involved in using them it would be the best of both worlds. Otherwise, "all you're doing is deskilling them rather than reskilling them", she said. She believes the tests were of good quality, but to be of value they needed to be marked by those teaching the course.
"It's counter-productive to have a test for literature at age 14 unless you're 100 per cent sure that 100 per cent of the teachers are current teachers of the age group in English."
Her team's report on external marking of the KS3 English tests says the Shakespeare paper was the most problematic.
The markers did not always know enough about the Bard and his use of language.
"Indeed one marker ticked one pupil's script when she wrote that Romeo uses 'Simples and Metfords'. Markers regularly ticked the use of appropriate terminology, even when the context suggested strongly that pupils did not understand either the terms or their literary purposes in the play," says the report.
It also says the brief training and limited samples provided to markers combined with the tight time-frame to result in markers working in their own way.
The mark scheme had originally been designed to be used in the pupils' own schools.
However, in a bid to lure teachers to give up the testing boycott of 1994, the Government agreed to hire external markers instead.
The Exeter report, highlights four significant trends: * a reluctance to use the full mark range; ndifficulties in discerning im-plicit understanding; * lack of pedagogic knowledge; * difficulties in assessing non-standard responses.
Evaluation of the Quality of External Marking of the 1995 Key Stage 3 Tests in English and Evaluation of Key Stage 3 Assessments for 1995 cost Pounds 5 each and are available from SCAA Publications, PO Box 235, Hayes, Middlesex UB3 1HF