National tests are a waste of time and money that yield little information about pupils not already known to their teachers.
Formal tests could be replaced by internal teacher assessments without affecting children's results, a study of more than 100,000 pupils found.
Two-thirds of pupils were given identical scores in their teacher assessments to those they achieved in formal tests, Danny Durant, a Worcestershire education adviser, told the British Educational Research Association conference, held last week in Edinburgh.
The system of national tests involves the biggest annual mailing in the country - 4.2 million test papers in 80,000 packages delivered over a two-week period - and costs the taxpayer more than pound;200m each year.
But Mr Durant said that the close relationship between test scores and teachers' views makes this costly process unnecessary. Only about one in 10 pupils had differences between their teacher assessment and test scores equivalent to a whole level. Five-year-olds were most likely to have large differences between the two scores, with 11-year-olds the least likely. The findings were based on Worcestershire pupils test scores between 1995 and 2002.
"The national curriculum assessment programme seems to have more to do with reporting results and politics than children in the classroom," Mr Durant said. "Teachers' assessments cost far less than the test programme and offer the opportunity for a more complete professional assessment of the students."
Meanwhile, a separate study presented to the conference found that it has become more difficult for 11-year-olds to gain a higher level in their writing tests.
The study by Sylvia Green, Martin Johnson, Nick O'Donovan and Pauline Sutton, of the University of Cambridge Local Examinations Syndicate, looked at the way key stage 2 writing tests were marked in 1995 and 2002.
They found that judgments remained similar at level 3 and level 4, the expected level for pupils leaving primary school. But markers now demanded a higher standard for level 5 than they had in 1995. The research was based on the writing tests of 396 pupils in 1995 and 540 pupils in 2002 from the same schools.
They found an overall improvement in schools' performance and concluded that it was genuine, adding: "Overall, children achieved higher levels in 2002 than in 1995 and not only did more children achieve level 5, but more sophisticated writing was expected at this level."