Speaking at a conference organised by the National Commission on Education to launch its concluding report, Ms Millett acknowledged that reduced class sizes could produce limited improvement in some circumstances.
But this is an expensive strategy, she told an audience of educationists at London University's Institute of Education, and is unlikely to be such good value for money as other methods.
"Reducing class sizes may be part of an overall strategy," she said. "But it is not an adequate form of intervention by itself. It is also expensive.
"We ought to be assessing its worth not in absolute terms but in relative terms."
She suggested that other approaches are liable to be more effective, including the sort of work carried out in the American "Success for All" project in Baltimore. This involves helping children with a number of educational strategies as soon as they are deemed to be at educational risk.
"We should not look at cutting class sizes and say simply 'yes, it's good' or 'no, it's bad'," she said. "There is a range of interventions we should make."
According to Professor Michael Bassey of Nottingham Trent University a reduction of the average 1993 class size of 27 to 24 would require an additional 26,000 teachers at an annual cost of Pounds 400-Pounds 600 million.
Class sizes have been a major bone of contention since they started to rise in the late 1980s. The issue was given particular prominence at the spring annual conferences of the teaching unions, all of which passed motions calling for reductions and limits on the number of pupils in a class. The National Union of Teachers balloted for industrial action in support of this (although the result was a vote against).
The National Commission's final report, formally published last week, recommended a reduction to 20 pupils in the first two years of primary school.
A recent survey by the National Foundation for Educational Research found that nearly 60 per cent of primary heads in England and Wales believed their key stage 2 classes were too big to enable "adequate" teaching of the national curriculum.
Class size was at the forefront of debate once more when, earlier this month, visiting researchers from Tennessee said that their experiment with smaller class sizes had led to significant improvements.
The Tennessee STAR (Student Teacher Achievement Ratio) project looked at evidence from 10 years in 79 primary schools. They found that children in classes of 15 did better than those in classes of 25. Tennessee has now passed a law limiting class sizes to 18 for five to eight-year-olds.