Government claims that class size is unimportant were called into question this week by research commissioned by heads, writes Clare Dean.
Christopher Day, from Nottingham University, said pupils in small classes misbehaved less, participated more, spent longer on their work and interrupted less frequently.
He also claimed that allowing class sizes to continue to rise was at odds with ministers' desire to improve pupil behaviour and standards.
The team, lead by Professor Day, was asked the NAHT to undertake an independent review of existing research.
Studies have found that pupils in small classes behave better, appear more absorbed in what they are doing, spend more time on task, and interrupt less frequently.
Interim findings showed that teachers with classes of 26 to 30 pupils on average worked three hours more per week than those with classes of under 21.
Professor Day said bigger class sizes led to more stress and added: "In the broad context, therefore, any savings made as a result of increasing class sizes will need to be offset by the cost of covering staff absence and turnover.
Last week the Government admitted that pupil:teacher ratios in primary schools this January were worse than a decade ago (23:1 in 1996 and 22.1:1 in 1986).
The Office for Standards in Education has claimed class size is linked to pupil achievement only in the first two years of primary education. And Gillian Shephard, the Education and Employment Secretary, has consistently argued there was no evidence to prove that larger class size led to poorer standards.
But David Hart, general secretary, said: "Anyone with an ounce of common sense knows that class size is a key factor in terms of the quality of pupil's learning.
* More than 40 per cent of primary pupils are taught in classes of 30 or more, according to new Government figures. The figures, based on 1995 returns, represent a 7 per cent increase on 1994. The number of secondary pupils in classes of 30 or more now stands at nearly 400,000.