A project by Birmingham education authority, one of the first of its kind, shows Caribbean boys scoring only a third of the male average in technical GCSEs.
The findings are supported by a major new analysis by the Association of Metropolitan Authorities and the National Foundation for Educational Research. This shows Caribbean pupils nationally are falling behind their white counterparts in maths and science.
Academic analysts are very concerned, pointing out that maths in particular is now crucial to progress in further education, training and employment. There is, for example, little chance of becoming a teacher without GCSE maths.
Britain's poor showing in the key areas of "number" and "communication" was a priority in Sir Ron Dearing's recent review of 16 to 19 education. He recommended that all pupils concentrate on them from the age of 14.
The figures have also led to renewed calls for the Government to introduce systematic ethnic monitoring at a local authority level. Ministers abandoned the last attempt in 1995, saying that the results were unreliable.
The NFERAMA study, an analysis of the 1994 GCSE results across 14 authorities, confirms the widespread view that girls are outperforming boys.
It also shows that most other ethnic groups - black African, Indian, Pakistani, Bangladeshi and Chinese - make more progress at school than indigenous whites.
Only black Caribbean pupils fare worse, particularly in maths where they score an average of one third of a grade below their white counterparts and in science where the difference is more than half a grade. They did appear, however, to have a slight advantage in English.
The Birmingham analysis, based on 1995 GCSE results, shows only 8.6 per cent of black Caribbean boys getting grades A to C in maths, and 12.4 in science. The comparable figures for other groups of boys are: black African - 14. 3 and 28.6; Indian - 34.2 and 44.1; and white - 32.2 and 36.9. Black Caribbean girls did considerably better, scoring 33.3 per cent A to C in maths and 37.5 per cent in science.
Another authority, which did not wish to be named, surveyed 1,508 pupils at 15 secondary schools and found that only 10 per cent of black Caribbean children got five grades A to C at GCSE (national average 43 per cent) and that only 7 per cent got any sort of GCSE pass in maths.
Birmingham produced its figures after working with the African-Caribbean Network for Science and Technology, a group of black scientists concerned about the failure of Britain's Caribbean pupils, boys in particular.
"What future will black Caribbean young people have in this Western, numerate, industrialised society?" asked the network's spokeswoman, chemical engineer, Liz Rasekoala. "They're not even qualified for an apprenticeship."
The network condemns the lack of detailed information. "Everybody was telling us there's a general problem. There was a lot of wringing of hands, moaning and groaning. But no one in the mainstream was willing to come out and say exactly what was going on.
"What we want from the Government is to have ethnic monitoring made statutory. As a result of gender monitoring there have been major improvements and support for girls. No one, apart from Birmingham, has had the moral courage to come forward with the figures."
Ms Rasekoala, who trained in Nigeria, said that black scientists are so rare they often feel isolated and may be tempted to seek other professions. She has spent her entire career as the only black individual at work and blamed low expectations at school for the failure of many black pupils in science.
Dr David Gillborn, a lecturer in policy studies at London University's Institute of Education, said that Birmingham's detailed figures are the first he has seen from a named local authority.
"Good quality ethnic monitoring is a crucial element in addressing existing inequality. But it must be with categories that are sufficiently detailed and meaningful." So far the Government is only proposing to collect very basic information.
John Fowler from the Association of Metropolitan Authorities said that ethnic monitoring was an important step, but that the central question of how to tackle underachievement still remains.