Grammar schools select the best 30 per cent or less, and overall, 95 per cent should get five top GCSE grades, Peter Downes, past president of the Secondary Heads Association, said.
But the Government's performance tables for this year, published yesterday, show that, though some grammar schools are at the top, 50 schools did not meet that target and 18 had scores of less than 90 per cent. At A- level 35 of the schools were below the national average.
The Grammar Schools Association contested Mr Downes' views as "offensive". It said grammar schools' results were better than ever.
Mr Downes, head of Hinchingbrooke comprehensive school, Cambridgeshire, said that with the Education Bill, which will extend selection, going through the Commons, ministers had to ask themselves whether some of the 161 grammar schools were failing.
"The national average for pupils getting five GCSE grades A-C is 44 per cent," he said. "Any grammar school that is not getting 95 per cent should ask itself why.
"Either the selection process is ineffective and therefore suspect or they are failing to teach children properly. One of the negative effects of selection is to depress the performance of those at the bottom."
Roger Hale, head of Caistor Grammar School in Lincolnshire, said the intake of grammar schools varied in different parts of the country. In some cities, the percentage selected was as low as seven while in other places it might be 34.
"As a headteacher I would not dismiss the performance of 50 schools without detailed information about their intake. We have been gathering information about test and exam results. Grammar schools are adding more value to their students than other types of school."
The publication of the tables caused controversy. Gillian Shephard, Secretary of State for Education, said: "Over five years tables have consistently driven up standards. This information provides an impetus for excellent schools and college to do even better, spurs those below average to set targets for improvement and galvanises poor performers into action."
However, David Hart, general secretary of the National Association of Head Teachers, said: "Any attempt to argue that the improvement has been largely the result of league tables would be insulting to the professionalism of school staff."
Critics say the tables do not present a fair picture of schools' performance because they do not take into account intakes. Labour has said that it will publish year-on-year comparisons to show how schools are improving and develop "value-added" tables to show schools' effect on pupils' progress.
Ministers are also investigating "value-added" tables.
t Research by Exeter University's geography department found that gender, school size and type and pupils' socio-economic backgrounds accounted for nearly 80 per cent of the variation in GCSE performance.Reuse content