A Commerce Department decision to change the method of counting military personnel stationed overseas in the 1990 census is expected to keep Wisconsin from losing a congressional seat and either Georgia or California from gaining a seat.

Commerce officials, who have jurisdiction over the census, had planned to count overseas military and their families as residents at the last U.S. military base where they had served for at least six months. That method of assigning residency would have boosted the populations of states with large military bases, including California, Texas, Georgia and Virginia.

States such as Wisconsin, where there are few military facilities, would likely have lost out in the population counts used to reapportion congressional seats after the census.

This method would have shifted a congressional seat from Wisconsin to either Georgia or California, according to an analysis by the Congressional Research Service.

But the Commerce Department recently announced it would instead count those overseas at their "home of record," the residence listed at the time the person joined the military. As a result, the numbers will be distributed more evenly across the country and probably will not affect the apportionment of congressional seats.

The "home of record" designation is distinct from "legal residence," declared by military personnel for purposes of paying taxes. That designation, which would have added substantial numbers to the populations of states with no income tax, such as Texas, was not considered seriously as a means of assigning residency.

The arcane process of placing overseas personnel at a U.S. residence has taken on significance because of the political stakes: There are between 1 million and 1.5 million persons living overseas, including non-military federal employees and their families, and even small differences in population can affect congressional reapportionment and the distribution of state legislative seats.

In 1970, when military personnel overseas were allocated to states using the "home of record," Oklahoma gained a seat that would have otherwise gone to Connecticut, according to congressional researchers.

"There really are big differences" in state population totals depending on the method of assigning residency, said Betty Mahoney, a staff member at the Defense Manpower Data Center. "The states which have large bases and relatively small populations are winners under 'last duty station' and losers under 'home of record,' " she said. The opposite is true for states with a small military presence.

Many members of Congress had objected to the initial plan for assigning residency because the Defense Department could provide information only on the last military base at which its personnel were stationed, not the state or locality in which they lived.

Persons assigned to the Pentagon before moving out of the country, for example, would have been counted as D.C. residents even if they had lived in Virginia or Maryland. The Pentagon is technically considered part of the District.

The House passed legislation late last month requiring the use of "home of record" to assign residency and similar legislation was pending in the Senate, putting pressure on Commerce officials to change their method of counting those overseas.

Rep. Thomas C. Sawyer (D-Ohio), who chairs the Post Office and Civil Service subcommittee on census and population, said he was "genuinely pleased" with the Commerce Department's decision. "Over a million Americans who are living abroad temporarily while serving our country will now be represented fairly in Congress," he said in a statement.

State population figures can vary significantly according to the different methods of assigning overseas personnel.

California, for example, would gain 110,729 residents using last duty station, but 54,741 using home of record, according to Pentagon figures. Georgia would gain 39,838 using last duty station and 21,363 using home of record.

Virginia would gain 54,479 under last duty station and 17,924 under home of record. Wisconsin would gain 319 under last duty station and 12,189 under home of record.

Military personnel outside the country were considered for reapportionment only once before, in the 1970 census. This will be the first time non-defense federal employees and their families will be counted.