Professor Alan Smithers, director of Brunel University's centre for education and employment research, said researchers had to become more rigorous if they were to continue to justify huge public expenditure.
In a devastating attack on his profession, delivered at the British Association's Annual Festival of Science at Leeds University, Professor Smithers said most research was either trivial, flawed, or biased.
Too many researchers were acting like social scientists, exploring esoteric issues instead of addressing the real issues of literacy and numeracy which would help teachers improve their performance.
And they had left a generation of children at the mercy of untested theories such as mixed-ability teaching which today seemed "daft in retrospect".
His speech adds to growing criticism of education research - Chris Woodhead, chief inspector of schools, last month called much of it irrelevant. And Dr David Moore, president of the association's education section as well as chief executive of the Association of Science Education,raised further doubts.
Addressing the same meeting, he called for a national audit of education research - "not just gathering the data but at the same time asking the people that should be using it what they want from it".
Professor Smithers' conclusion is that most education research is not wanted. Despite receiving some #163;100 million in public funding each year, the results are largely ignored by everybody from politicians to classroom teachers.
And he highlighted a string of papers published this year which were, in his view, "gobbledygook", including "Between Postmodernism and Nowhere: the predicament of the postmodernist".
"This isn't going to help you to teach reading and subtraction.
"The way forward is to be clear about what research can and cannot do. It can tell you what is, never what ought to be." Decisions about the best way forward were for others to take, he said.
Professor Smithers admitted that education research was difficult because of the impossibility of running controlled experiments. Unlike science, which examined a constant reality, education was determined by policy - and you could never persuade people the educational world was round if they believed it was flat.
Politicians had also proved remarkably adept at ignoring evidence that did not suit them, but that was no excuse for ducking the issues. Unfortunately, the response of many researchers had been to attempt trivial questions or simply to forget their role was to find ways of improving education. Instead they adopted the role of social scientists studying education at arm's length.
"Education is an applied area. We want to improve the quality of that practical activity, not discover some-thing new about the earth," he said.
That meant tackling problems in literacy and numeracy head on: identifying the routine errors children make, finding ways to diagnose them and devising effective methods of correcting them.
Otherwise, he warned, education would continue to be at the mercy of the whims and follies of politicians and others with an agenda to pursue. That had happened with strategies like mixed-ability teaching and leaving the youngest children to pick things up for themselves.
"Which of today's other crazes will be shown to be the educational equivalent of UFOs - much believed-in but with no hard evidence to support their actuality?"
Philippa Cordingley, chief professional research adviser to the Teacher Training Agency - who reminded the audience that Professor Smithers had once described it as having a "pathetic attempt at a research policy" - was more optimistic. She argued the key to successful research was to involve teachers, and said the TTA was trying to make research more relevant and accessible.
"The research teachers are looking for talks about the detail of what happens in classrooms. Of course, a lot of research won't start there but it has to beat the path back there."