National benchmarking data, issued to allow schools to set their own targets, will group schools according to the number of pupils taking free school meals and whether a majority of pupils speak English as a second language.
In effect, that will allow schools in deprived areas to set more realistic targets. They will find their performance compared with schools in similarly deprived boroughs.
The data will also show the results of the best and the worst of the schools in each group, underlining the Government's belief that poverty is not a barrier to success.
And from 1999, schools will also be able to use "prior attainment" data - matching targets to the results obtained by pupils at the previous key stage.
The new proposals are outlined in Target Setting and Benchmarking in Schools, a consultation paper issued this week by the School Curriculum and Assessment Authority.
It sets out a timetable for schools to set and meet their own targets. Targets for summer 2000 must be in place by Christmas 1998. But the first national data will be published in November, allowing schools to pilot the system this year.
Analysis of maths tests taken by 11-year-olds in 1996 revealed a 20 percentage point difference in median scores between schools with less than 10 per cent of pupils on free meals and those with more than 35 per cent.
SCAA proposes six bands, covering differing uptake of free school meals, and a seventh for schools where more than half the pupils speak English as a second language.
Although the Government has promised an additional Pounds 1 billion for school budgets, this is unlikely to do more than stave off cuts .
The Government is planning 20 education action zones which will have first call on funding for school improvement initiatives - though no new cash from the Treasury.
National Union of Teachers education officer John Bangs said: "There is a real problem with action zones. The concept has not been thought through."
Peter Jackson of SCAA said that many heads wanted to see schools banded by socio-economic factors, even after prior attainment data became available.