President Clinton and veterans from across the nation will gather in somber remembrance this afternoon at the Korean War Veterans Memorial, where it is carved into stone that 54,246 Americans died in the war that began 50 years ago today.

For decades, the number has been enshrined in almanacs, histories, memories and monuments, cited as proof of the war's cost. But nearly one-third of those deaths--17,730--occurred elsewhere, often half a world away from Korea, in places ranging from the United States to Germany. The actual number of Americans killed in the Korean War theater of operations is 36,516, the Pentagon acknowledged this month.

"If you were walking down the street in Washington, D.C., and were hit by a car, you'd be considered a casualty of the Korean War," said Burt Hagelin, a Korean War veteran who helped uncover the mystery.

The error was blamed on an anonymous government clerk who in the 1950s mistakenly added all noncombat deaths worldwide to the total, and the correction was credited to revised reporting procedures, according to news accounts reporting the Pentagon's clarification. That is not the real story, according to veterans and others who have pushed for years to correct the number.

"Fifty years later, they're trying to drop it all on one clerk," said Richard Kolb, publisher of VFW Magazine, a Veterans of Foreign Wars publication that several times has urged the number be corrected. "They had the facts all along. Now they're acting like it's a new revelation."

Some Pentagon officials have for years considered the 54,000 figure inflated, and they believed before the memorial was dedicated in 1995 that engraving that number in black granite would be misleading, according to interviews. But at the insistence of the veterans committee that oversaw the memorial's construction, the figure was used.

Far more than a numbers game, the issue is emotional to many Korean War veterans, who see the revised total as another slap at their oft-ignored sacrifices.

Some of the veterans, tired of their conflict being slighted in favor of the Vietnam War, say 54,000 is accurate because all those killed were part of the war effort, regardless of where they died. Veterans are angry at suggestions from VFW Magazine and others that the number engraved along the memorial's Pool of Remembrance be changed.

"We're committed to the total cost of the war: 54,000," said retired Marine Gen. Ray Davis, who was awarded the Medal of Honor after the Marine breakout from the Chosin Reservoir and who was chairman of the Korean War Veterans Memorial Advisory Board when the memorial was dedicated. "Under no circumstances would I like to see the memorial changed."

If the Vietnam Veterans Memorial included out-of-area deaths along with the 58,000-plus names engraved on the wall, it would gain more than 20,000 names of Americans who died in the United States and elsewhere from 1965 through 1975, according to a study cited by VFW Magazine.

Ironically, two Korean War veterans from Maine are responsible for bringing the issue to the fore through years of prodding, according to Kolb.

Hagelin, who was with the 2nd Infantry Division during the bloody fighting for the outpost known as Old Baldy, read an article in the early 1990s that said the 54,000 Americans killed in the Korean War included 20,000 non-battle deaths. "I couldn't believe it," said Hagelin, 69, a self-described "Maine farm boy" living in the town of Dover Foxcraft. "I didn't believe the percent of non-battle deaths could be correct."

Hagelin started making inquiries at the Pentagon, asking to see statistics on the non-battle deaths, which usually stem from accidents, training fatalities or disease. "I'd get stonewalled," he said. "They kept no records on it, they claimed."

But Hagelin obtained some data after members of Maine's congressional delegation applied pressure on his behalf. He teamed up with a former soldier from Augusta, Marty O'Brien, who was independently investigating the matter. "The numbers were all over the place," said O'Brien, 70, a 1st Cavalry Division veteran.

Through the Freedom of Information Act, O'Brien obtained a microfiche showing the Army's non-battle deaths during the Korean War, information the Army had claimed did not exist. Working together, Hagelin and O'Brien helped establish that most of the 20,000 non-battle deaths had occurred outside Korea.

Following their efforts, the Pentagon published statistics in 1994 and subsequent years showing that 17,000 of the non-battle deaths had occurred elsewhere. About 2,800 of the non-battle deaths had occurred in the theater. When that figure was added to the 33,600 killed in battle, the number of Americans who died in the war came to fewer than 37,000.

Nonetheless, the Pentagon made no effort to publicize the new number, and the 54,000 figure continued to be generally accepted, particularly after the Korean War memorial opened.

"I had been there and seen the number 54,000 on the stone, so I believed it," said retired Air Force Maj. Gen. Nels Running, director of the Pentagon's Korean War Commemorations Committee and a former senior commander of U.S. forces in Korea.

The committee used the higher figure in promotional bookmarks it published early this year. "The 54,000 number became so pervasive in the secondary literature that we felt it not necessary to go to the primary sources," said Air Force Maj. Bob White, the committee's historian.

White said he learned that the figure was suspect from a military historian who saw one of the bookmarks. "He said, 'You have a problem.' And indeed we did," White said.

The 54,000 figure dates to the war's end in 1953, reflecting about 33,600 killed in battle and 20,600 non-battle deaths worldwide.

The error appears to have resulted from an effort to make the Korean War figure more accurate.

In the 1950s, the Pentagon focused on battle deaths, said Thomas Campbell, an official with the Pentagon's Directorate for Information Operations and Reports, which is responsible for casualty figures. So, the figure of roughly 33,000 was generally used to describe Korean War deaths, rather than 54,000.

During Vietnam, the Pentagon included non-battle deaths in reporting casualties in that war, Campbell said. Officials then decided that for consistency, Korea's non-battle fatalities should be added to the war's death toll in the annual almanac of defense statistics. "That's when the confusion came into play," Campbell said.

The only non-battle deaths included for Vietnam were those in the theater. But all non-battle deaths were added to the Korean figure, without any explanation that the number represented a worldwide total.

The total of 54,000 became the common reference for Korean War deaths. For many Korean veterans, the fact that it is close to the 58,000 killed in Vietnam is significant. "It's a matter of personal pride for them: 'We had a more brutal war. We lost almost as many in three years as you lost in 10,' " Running said.

According to Campbell, "nothing much was said [about the out-of-area deaths] until the Korean memorial went up. The 54,000 is the way that the people who built the memorial wanted it, whereas the [Pentagon] view was to treat it the same way as Vietnam."

The veterans' view won out. "Had it not been for the Korean War, they would never have been drafted," Bob Hansen, executive director of the Korean War Veterans Memorial Advisory Board, said of those who died. "It was the shake of the dice where they were sent."

Kolb said those who died elsewhere during the years of Korea and Vietnam would be better honored on a Cold War monument.

The Korean War had plenty of death as it was, he said. "Almost 37,000 Americans killed in three years is more than enough."