Next year's census could give 10 more seats in Congress to the South and West as those regions draw population away from the shrinking Midwest and Northeast, but change will be less dramatic than a decade ago, Democratic and Republican analyses show.

The two studies said it appears that 15 states would win or lose seats in the House of Representatives, based on current population trends. After the 1990 census, 21 states gained or lost seats.

The seven winner states are all in the South or West--Arizona, Texas and Georgia, which would gain two seats each, and California, Colorado, Florida and Nevada, which would gain one seat each. Most of the eight loser states are in the Midwest or Northeast--New York and Pennsylvania, which would lose two seats each, and Connecticut, Illinois, Mississippi, Ohio, Oklahoma and Wisconsin, which would lose one seat each. The trends are consistent with population shifts that have been occurring for decades.

The tallies are included in reports by Kimball W. Brace of Election Data Services, a Democratic consulting firm in Washington, and by Clark Bensen of Polidata, a Republican consultancy in Northern Virginia. Both men added one new winner state--Colorado--and one new loser state--Illinois--to their tallies from reports they did last year.

Their conclusions are based on state population estimates released yesterday by the Census Bureau, showing that the fastest-growing states are in the West and South, and the Northeast and Midwest the slowest-growing regions.

These are the final population estimates before the 2000 census, which will redistribute congressional seats among states. Every state gets at least one seat in the 435-member House, and the remaining 385 are divided up using a complicated formula based on population and growth, with the intent of creating districts equal in size.

The older industrial Northeast and Midwest has been losing people for years to the South and West, and Republicans do well in those growing regions. The GOP also dominates at least half the state governments, compared with two in 1990, Bensen said. State legislatures, which must redraw boundaries in time for the 2002 elections, have considerable leeway to do so.

"The Republicans have not been in as good shape for redistricting since 1930, in terms of being able to control the outcome within the states," said Rep. Thomas M. Davis III (R-Va.), head of the Republican Congressional Campaign Committee. "The one variable is who controls the process."

That and other factors, experts say, make it difficult to predict if the census will only help Republicans. For one thing, California--the biggest prize, with a potential 53 seats--is controlled by Democrats. Elections could change the balance of power in some states.

Some states could boost their census totals, and congressional representation, by staging aggressive be-counted campaigns.

In a victory for Republicans, the Supreme Court ruled that only a door-to-door count could be used to reapportion congressional seats, and barred use of statistical sampling that is intended to compensate for an undercount.

But some legislators hope to use sampling, which tends to help Democrats because it adds to the count of minorities and poor people, to redraw congressional and other boundaries within states--a move Republicans promise to challenge.

Montana's congressional fate is up in the air, both consultants said. Based on the 1999 estimates, it would barely win back the second seat in Congress that it lost after the 1990 census. But when the consultants used current trends to project those estimates ahead to census day next April, Montana would not gain.

One striking feature of this past decade, Brace said, is how the pace of change has slowed. "Everything is growing much more evenly than the dramatic shifts like there were before," he said. "In 1990, California gained seven seats. Now, they are only taking one. The highest of any state is two seats."

The census population estimates, which are based on government records such as birth and death certificates, may not match the census count, Bensen and Brace said. In addition, the census includes military and federal employees working overseas, and credits them to their home states, which the estimates do not.