The first ever stats that show pupils' performance by ethnicity at the beginning of compulsory education have been released.
Figures from the Department for Children, Schools and Families make it possible to compare the achievement of ethnic groups at the age of five to later on in school life.
Pupils with a mixed white and Asian background (56 per cent) and the Irish (53 per cent) were most likely to have a good level of development at five, followed by white British, Indian and Chinese. But by seven, Chinese and Indian pupils have overtaken white British pupils in reading, writing and maths.
Just one in three Bangladeshi pupils had a good level of development at five in 2008, the statistics show, compared with 49 per cent of pupils overall.
Bangladeshi pupils' performance at seven is similar to that at age five. In 2008, one in three failed to reach level 2b in reading - the "secure" level at age seven. But they had improved relative to other groups. Black African, black Caribbean, Pakistani and white children who are not British or Irish did worse.
All the figures are based on results from the Early Years Foundation profile assessments being carried out in the classroom.
Dr Abdul-Hayee Murshad, head of Hermitage Primary in Wapping, east London, where 80 per cent of pupils are Bangladeshi and results at age 11 are above the national average, said: "This underlines the fact all children do not have the same starting point. That is why some make good progress and do not meet the threshold, but others meet the threshold but have not made good progress.
The foundation stage profile assessment is nothing to do with education or children. It is just a political measurement, it doesn't link to key stage 1 and local authorities set the targets even though it is the schools who know the children."
Traveller, Gypsy and Roma children are recorded as the lowest achieving at both ages, with less than one in five having a good level of development at age five.
However, the DCSF warns that these figures may be distorted because many pupils are not present for either the annual schools census or the later assessments on which the statistics are based.
Much of the difference in children's scores is linked to deprivation. Only 31 per cent of children on free school meals achieve a good level of development compared to 52 per cent of those not eligible. Bangladeshi children are more likely to claim free school meals and less likely to have English as their first language than other groups.
A breakdown by areas within local authorities show that Tynedale in Northumberland had the highest scores with an average of 72 per cent of five-year-olds rated as having a good level of development compared to 28 per cent in Barnsley, South Yorkshire. But this wide range may also indicate the difficulties in moderating assessments nationwide.
But previous research has shown that the differences between ethnic groups cannot be entirely explained by their relative wealth.