Ken Boston, chief executive of the Qualifications and Curriculum Authority, said the UK should follow the lead of Far Eastern nations in doing much more to challenge clever pupils.
Korea, Singapore, Japan and Hong Kong were "well ahead" of the UK in international comparisons of pupils' performance in maths and science, he said.
In those countries, he said: "There is a different and demanding curriculum for the higher-ability groups, tested by different and more demanding assessments, shaped and taken only in response to their needs."
This is one of the foundations of their success, he told a London conference to launch the QCA's annual report. The idea puts Dr Boston at odds with much of the philosophy behind the GCSE and A-level, which has emphasised access to common teaching and exams for pupils of all abilities.
But he used the comparison with the East to highlight work the Government has already announced in this field.
The QCA is investigating the introduction of harder A-levels from 2009, possibly using A*, A** and A*** grades to recognise achievements above an A grade.
This month, Ruth Kelly, the Education Secretary, announced plans for a new GCSE in further maths, targeted at high achievers.
Dr Boston said that in future A-levels would put more emphasis on judging pupils' extended essay-writing and reasoning skills but said this did not amount to "elitism". There was no trade-off between the needs of higher- and lower-achieving pupils, he said.
He said the Government's drive to create a new vocational or "specialised"
diploma from 2008 was "the most ambitious and important educational reform currently being undertaken anywhere in the world".
But he said it carried risks. Pupils would have to be taught by "industry-qualified" staff, and in industrial-standard facilities.
In response to a QCA invitation to set out priorities for the future, the Mathematical Association said this week that frequent testing in schools was damaging learning.
It said that tests encouraged coaching that focused solely on the "standard questions that appear on papers and leads to the exclusion of more interesting and challenging problems at all levels".