"He has the incumbent's curse on a number of issues," said Peter Feaver, a political science professor at Duke University in North Carolina and an authority on US civil-military relations, referring to the disappointment over unrealized campaign promises.

Bush is believed to have won the military vote hands down against former vice president Al Gore in the 2000 election.

Absentee ballots cast by military deployed overseas helped push Bush over the top in Florida in 2000, and his 537-vote margin of victory there ultimately gave him the presidency.

But four years later, the excitement Bush once generated with promises to rescue an overworked military has been offset by events.

Wars in Afghanistan and Iraq have put active duty and reserve military under unprecedented stress with frequent, lengthy and dangerous deployments. In Iraq alone, the death toll has surpassed 1,000.

Defense Secretary Donald Rumsfeld and other Republican appointees had stormy relations with the military brass, including clashes over the size of the army and other issues.

Some retired generals who sided publicly with Bush in 2000 after serving under president Bill Clinton have switched sides, this time lending vociferous support for Senator John Kerry, the Democratic challenger.

How much of an impact all this will have on service members in November is still a matter of guesswork.

Little polling data exists on how the military actually voted in 2000, although it is widely assumed it went for Bush by a large margin. Exit polls generally have not been performed at military bases where service members cast their ballots.

Public opinion polling of the active duty military also has been scant because the 1.4 million member force represents a tiny constituency that in itself is unlikely to alter the election outcome even in battleground states.

Polls of veterans, however, show Bush doing better than Kerry among Americans with military experience.

A Pew Research Center poll released on July 23 found that Bush led Kerry 49 to 40 percent among male veterans, a group that included active duty military personnel. Non veterans were more evenly divided -- 46 percent for Bush versus 44 percent for Kerry.

Nevertheless, students of the military detect a shift in sentiment when compared to the last presidential elections.

"You don't find much antipathy toward Bush in the way that there was antipathy toward Clinton. But there is disenchantment in some quarters," Feaver told AFP.

"The military support the Iraq war much more strongly than the general public does. But there is still some ambivalence among the military, and it is plausible that that ambivalence is growing as the situation in Iraq proves so difficult," he said.

Bush may be particularly vulnerable among the national guard and reserve and their families, who have endured longer, more dangerous active duty deployments than anticipated, he said.

Before the 2000 election, Feaver had tracked a growing tendency on the part of US military officers to identify with the Republican party.

"Republicanization (of the military) has receded a bit," he said. "It hasn't flipped to loyalty to the Democrats. It probably has increased the number of people who would identify themselves as independents."

But it was still clear Bush would win the military vote, he said.

Kerry, he said, "has not presented such a compelling person, or platform or campaign that would cause a constituency that was a Bush stronghold to completely reverse.

"He might do better than Gore did, and in a very, very close election that might matter. But there are very few states where the military vote is large enough, where the election will be close enough, where Kerry's eating into Bush support will be decisive," he said.