As the nation debates how to reform the intelligence community, people should take a careful look at one of Washington's hidden jewels -- the State Department's tiny Bureau of Intelligence and Research.

Despite its small size (or more likely because of it), the bureau has what many regard as the best track record in the government in assessing intelligence issues for policymakers. In the recent Iraq debates, for example, it was the only intelligence agency that was consistently skeptical about whether Iraq had chemical or biological weapons. And it warned before the March 2003 invasion about the political and ethnic turmoil that was likely in postwar Iraq.

"INR," as State's intelligence bureau is known in the government, has just 305 analysts. That's a fifth of the 1,500-plus at the CIA's Directorate of Intelligence and about a tenth of the 3,000 or so at the Pentagon's Defense Intelligence Agency.

Many proposals for reforming the intelligence community after its failures concerning the attacks of Sept. 11, 2001, and Iraq have argued that in the spy world, bigger is better: more people, broader responsibility, greater interagency coordination.

But INR's success story suggests that small is sometimes beautiful. Because it is little, INR tries to maintain an elite reputation. And because it is intimately connected with State Department policymakers, it never loses sight of what the consumers of intelligence actually want: sound judgment.

INR is that rare agency that has been shrinking over time. Its workforce has dropped from more than 1,600 people in 1945, when it inherited the responsibilities of the wartime Office of Strategic Services, to a bit over 300 by 1961. It has stayed at roughly that level ever since -- a bureaucratic "steady state" that is almost unheard of in the nation's capital.

INR is the "biggest little intelligence shop in town, where they really do a lot more with a lot less," argued the late Les Aspin, who was defense secretary in the Clinton administration.

What the State Department bureau lacks in numbers it makes up in expertise. The average analyst has 11 years of experience in his area of expertise, four times as long as the CIA average, according to a State Department official.

Many INR veterans have several decades of experience in their areas of specialization. The Near East South Asia section chief has been analyzing that area for 25 years; the European chief has spent 24 years studying his region. And because the bureau is so small, each analyst has broad responsibility; one person covers all the German-speaking countries in Europe; another has responsibility for all the Scandinavian countries.

The reason INR has been so effective, State Department officials say, is that it has maintained a culture that supports dissent -- and demands expertise. "We'd rather be right than quick," says one State Department official. Within the intelligence community, the State Department's analysts say they are seen as "malcontents" who demand hard evidence before they sign off on estimates. Their style is to ask: "What are the facts? How much do we know? Does the evidence all point in one direction?"

Over the years this culture of skepticism has produced some famous successes: INR provided more accurate bomb damage assessments during the Vietnam War than did the Pentagon; it warned in the late 1970s that if the Carter administration allowed the deposed shah of Iran to enter the United States for medical treatment, there would be trouble in Tehran (in the end the U.S. Embassy was seized).

In the Balkans, the bureau correctly cautioned that a bombing campaign would not force Serbian leader Slobodan Milosevic to leave Kosovo quickly. And on Cyprus, INR continues to warn that the elements of a Greek-Turkish settlement aren't yet in place.

But it's on Iraq, where the Bush administration's rush to judgment proved so costly, that INR has distinguished itself. A year ago it criticized the administration's theory that Iraq would be the beginning of a pro-democracy toppling of dominoes in the Arab world. It warned that Turkey would feel threatened enough by the prospect of Kurdish autonomy that it might not allow U.S. troops to transit its borders into Iraq.

What INR has, above all, is a culture that rejects easy answers and shoddy work. A State Department official remembers his early years in the bureau, when reports would be sent back to him full of corrections and notations such as, "Start over" or "You missed it." When that kind of intolerance for mediocrity is shared throughout the intelligence community, we'll know that reform has really begun.