Ask Americans if they favor the principle of civilian control of the military, and the vast majority would say yes. Ask if they favor politicians meddling with the conduct of military operations, and the vast majority would surely say no. But ask where the line is drawn between legitimate civilian control and unwarranted political interference, and, I suspect, the response would be less clear-cut. Many Americans' answers to that question would probably not reflect a considered philosophical viewpoint but an opinion about the particular military operation--and the particular political leadership--in the case at hand.

The issue made its most recent reappearance with the news that U.S. Army Gen. Wesley K. Clark, supreme commander of NATO, will be relieved of that post earlier than had been expected. Commentators widely assumed Clark's early exit was ordered because of his disagreements with civilian leaders over how to conduct NATO's 11-week campaign to force Serbian troops out of Kosovo.

Most critics found fault with the civilians, not with Clark. Presidential contender Sen. John McCain (R-Ariz.), for example, voiced concern that Clark's early departure "might influence other senior officers to place political considerations before military necessities and deprive future presidents of the counsel they need to best protect our security."

This is hardly the first time politicians have been accused of making military success more difficult--or even impossible. Gen. George B. McClellan during the Civil War and Gen. Douglas MacArthur during the Korean War both expected to run for president on exactly that issue. Gen. Joseph W. "Vinegar Joe" Stilwell, who commanded the China-Burma-India theater in World War II, considered President Franklin D. Roosevelt a "rank amateur in all military matters," prone to "whims, fancy and sudden childish notions" about waging war. Adm. Ulysses Grant Sharp, commander of Pacific forces during the Vietnam War, complained that the Johnson administration had forced the armed forces to fight "with one hand tied behind their backs"--a view widely espoused by other commanders of the Vietnam era.

Large numbers of people agreed with these military men, and it would not be hard to find cases in which politicians' ignorance, inexperience, hubris or lack of moral courage led to military disasters. Winston Churchill, usually held up as the model wartime leader, might qualify for some sort of grand prize in this category, having disregarded the counsel of his best military advisers about Greece and Singapore, thus bringing about the two worst British catastrophes of World War II within several months of each other.

Yet if we believe in civilian control, how can we complain? If the phrase means anything at all, it means that politicians and political appointees will have the final say in military matters.

Many people have little trouble resolving this paradox. Let the civilians make the basic decisions about when to go to war, against whom and for what objectives, and then step aside and let the military do its thing. The great 19th-century German general Helmuth von Moltke liked this approach: Once war begins, "political considerations can be taken into account only as long as they do not make demands that are militarily improper or impossible."

The problem with this argument is that wars are fought for political considerations, as Karl von Clausewitz underlined in his well-known observation that "war is merely the continuation of policy by other means . . . . Therefore there can be no question of a purely military evaluation of a great strategic issue, nor of a purely military scheme to solve it."

Soldiers generally know a lot more about conducting military operations than civilians. But all military operations have political consequences. That was one reason the war with Serbia had to be conducted as it was. The political aim of keeping the NATO allies together seemed to demand it.

The most controversial decisions--the incremental pace of the bombing, the question of committing ground troops to Kosovo and the decision to attack the economic infrastructure in heavily populated areas--all had important political and psychological implications far beyond the question of whether destroying an electric power grid would help bring about Yugoslav President Slobodan Milosevic's capitulation. These wider implications are properly the responsibility of elected political leaders and not professional military officers.

So how far should civilians go in micromanaging military operations? Lyndon Johnson's direction of the Vietnam War has been held up for almost three decades as a prime example of what not to do. Johnson and his civilian advisers exercised minute control over air operations against North Vietnam. Their methods have been widely condemned for needlessly endangering U.S. pilots' lives and hobbling the air war. (It is often overlooked now, but the primary reason Johnson kept such tight control was not a tenderhearted concern about civilian casualties or worry about the reaction of antiwar protesters, but a dread of inadvertently provoking a crisis with Russia or China. If LBJ has access to CNN wherever he is now, he must have had a knowing grin when he saw the repercussions of the U.S. bombing of the Chinese Embassy in Belgrade.)

Soldiers often dislike political control because in their eyes it makes their missions more dangerous and costly, sometimes seemingly impossible. Or they think they are being asked to perform tasks that the armed forces should not do at all--as in the ill-fated U.S. intervention in Somalia earlier in this decade. But we can be fairly certain that political leaders in the future will again assign the military services to inappropriate, costly or militarily dubious missions. This does not occur simply because of ignorance or whimsy on the part of the politicians. Rather, it reflects the fact that politicians must answer to a far wider range of demands than do soldiers. Along with military considerations, a president may have to be concerned with the attitude of his allies, the reaction of his domestic constituency, the state of the economy, the larger goals of his foreign policy and the "signals" he is sending to foreign rivals and potential enemies.

In 1942, FDR rejected the advice of his chiefs of staff to push for an early troop buildup in Britain to enable the Allies to launch an invasion of Northern Europe in 1943. Instead, the president went along with Churchill's proposal for a 1942 invasion of North Africa. That decision, in effect, killed any possibility of an invasion of France in 1943. It was not that Roosevelt failed to understand the advice of his military chiefs, but he knew that the public expected American troops to be fighting the Germans somewhere a year after the United States had entered the war. As Gen. George C. Marshall later told his biographer, Roosevelt and Churchill were dealing with political necessities--"something that we [military officers] fail to take into consideration."

So what is a military leader to do when faced with an ill-advised or unworkable mission? He could resign. The top military leadership of the Vietnam era has come in for a lot of after-the-fact criticism for not choosing that option. Senior commanders knew the war was being misdirected, critics have said, and they should have resigned rather than carry out a flawed policy.

This is an implausible scenario. If career officers were to quit the first time they were given a harebrained assignment, they would all leave as lieutenants. Clark did the right thing as commander of NATO's Kosovo campaign: He did his best to persuade his civilian superiors and his allies that they ought to follow his advice, and when they refused to do so he carried out the mission as best he could.

Critics of political meddling like to quote MacArthur's famous line that in war there is no substitute for victory. In fact, there are a lot of substitutes for victory, and soldiers are sometimes asked to bring them about--such as holding the line while national leaders search for a compromise, for example. Following the publication of my book "After Tet," about the two worst years for Americans in Vietnam, I was invited, possibly by mistake, to talk to a reunion of Army Rangers. After listening in glum silence to my depressing summary of the military picture of that stage of the war, one man finally asked, "You mean we were just there to buy time?" "Yes," I replied. "That's what soldiers often have to do."

Like freedom of speech, free enterprise, equal justice and other American ideals, civilian control of the military will continue to be both universally valued and universally debated. Soldiers and politicians will never have the same responsibilities or concerns, or the same vision of the world and its conflicts, and some measure of tension and misunderstanding will probably always exist in their relationship. But the continuing debate should not obscure the fact that for more than 200 years, civilian control of the military has achieved its central purpose. No U.S. military leader has used the armed forces to defy the processes of democratic government; no U.S. president has feared a military coup or has had to doubt that those in uniform will follow his direction--whether they think it wise or not.

Ronald Spector is a professor of history and international relations at George Washington University and the author of "After Tet: The Bloodiest Year in Vietnam" (Vintage).