OVER THE YEARS, public views of the Central Intelligence Agency and its role in American foreign policy have been shaped primarily by movies, television, novels, newspapers, books by journalists, headlines growing out of congressional inquiries, exposes by former intelligence officers and essays by "experts" who either have never served in American intelligence, or have served and still not understood its role.

The CIA is said to be an "invisible government," yet it is the most visible, most externally scrutinized and most publicized intelligence service in the world. While the CIA sometimes is able to refute publicly allegations and criticism, usually it must remain silent. The result is a contradictory melange of images of the CIA and very little understanding of its real role in American government.

A review of the CIA's basic functions -- intelligence collection, covert action and providing analysis for policymakers -- yields a textbook description of the role of intelligence. This textbook approach is neat, unambiguous, clinical, noncontroversial, even commendable -- and highly misleading.

It does not address central questions such as whether certain users of intelligence seek not data or understanding but support for decisions already made; whether they selectively use or misstate intelligence to influence public debate over policy; whether they disingenuously label intelligence they dislike as too soft, too hard or "cooked"; whether some intelligence officers are addressing personal agenda or biases. It does not reveal the implications for intelligence and policy of a CIA director who is held at too great a distance from the president, or of one who is too closely associated. It does not treat policymakers' frustration with constantly changing evaluations or with analysis that is just plain wrong, or the use of intelligence as a political football in struggles between government departments or between the executive and legislative branches.

The attitudes and actions of CIA officials and policymakers that lie behind these and many similar issues, and the interaction among them, comprise the dynamic of the relationship -- what Prof. Yehoshafat Harkabi of the Hebrew University of Jerusalem describes as "the intelligence-policymaker tangle."

The fact is that, over the years, the policymaker and the intelligence officer have consistently (and with frighteningly few exceptions) come together hugely ignorant of the realities and complexities of each other's worlds -- process, technique, form and culture. CIA officers can describe in excruciating detail how foreign policy is made in every country in the world save one -- the United States. By the same token, intelligence officers are troubled when they compare the precision of their collection and analysis against the sometimes cavalier formulation of policy.

The policymaker has a long list of grievances, many legitimate, some not. Policymakers legitimately want intelligence information that will inform and guide their tactical day-to-day decisions. In some areas, the CIA can and does meet their needs. For example, in 1980, thanks to a very brave man, the agency was able to provide policymakers with knowledge of the step-by-step preparations for the imposition of martial law in Poland. In early 1986 it was able to document, in extraordinary detail, electoral cheating in the Philippines. There are even some areas where CIA intelligence is so good that it reduces policymakers' flexibility and room for maneuver.

Yet I would have to acknowledge that there are countries and issues important to the United States for which tactical intelligence remains sorely deficient, and here the complaints of policymakers are justified. The CIA's capabilities have improved much in recent years, but they are still uneven in quality. And no matter how good CIA intelligence is, there will still be surprises or gaps.

It is no surprise that few policymakers welcome CIA information or analysis that challenges the adequacy of their chosen policies or the accuracy of their pronouncements. Indeed, during the Vietnam War, this was a constant issue. On the other hand, I concede that on more than a few occasions, policymakers have analyzed or forecast developments better than intelligence analysts.

And, truth be known, analysts have sometimes gone overboard to prove a policymaker wrong. When Secretary of State Alexander Haig asserted that the Soviets were behind international terrorism, intelligence analysts initially set out, not to address the issue in all its aspects, but rather to prove the secretary wrong -- to prove simply that the Soviets do not orchestrate all international terrorism. But in so doing they went too far themselves and failed in early drafts to describe extensive and well-documented indirect Soviet support for terrorist groups and their sponsors.

Far from kowtowing to policymakers, there is sometimes a strong impulse on the part of intelligence officers to show that a policy or decision is misguided or wrong, to poke an analytical finger in the policy eye. Policymakers know this and understandably resent it. To protect the independence of the analysts while keeping such impulses in check is one of the toughest jobs of intelligence agency managers.

Many policymakers believe the CIA allows its biases to dominate its reporting. Who would disagree that CIA officers have views and biases, and that they try to promote them? But the CIA is not monolithic; there is a wide range of views on virtually every issue. Indeed, internal debates are fierce and sometimes brutal -- after all, the stakes are very high. A classic example was the debate over the validity of the Sino-Soviet split in the early 1960s. A more recent one was the bitter internal disageement over who was behind the attempted assassination of Pope John Paul II.

More serious is the accusation that an institutional bias affects the CIA's work. I believe there probably is bias in some areas, in a broad sense. Although the agency was severely criticized for underestimating Soviet missile deployments in the late 1960s and early 1970s, it is probably regarded today as more skeptical than most observers of Soviet intentions, as more cynical about the public posture of other governments when contrasted to their overt and covert actions, as more doubtful about the ease and speed with which the United States can usually affect developments overseas, and, fairly consistently, as tending to see perils and difficulty where others do not.

Suspicion that the CIA's assessments are biased is especially great in areas where the CIA is involved in covert action. But such suspicion is greatly overdrawn. Analysis and operations are carried out by two completely separate directorates within the CIA. There are long-standing cultural as well as bureaucratic differences between the two branches. There is -- unfortunately, in my view -- little flow of people serving in one directorate to positions in the other. Thus the analysis of developments in countries where covert activities are under way is done by people with no role or stake in the operations. A common source of tension is an analysis that is far less optimistic than the directorate of operations' people would wish -- but, by policy, the analysis stands.

The impatience of policymakers with with the CIA is intensified by the fact that analyses and forecasts, often based on incomplete or ambiguous information, are sometimes wrong, and often change or are revised to take account of new information or new analytical techniques. CIA assessments of Warsaw Pact strength have changed on several occasions, complicating the task of U.S. negotiators in the Mutual and Balanced Force Reductions talks in Vienna. Similarly, the agency often revises its assessments of Soviet strategic weapons as new information becomes available. The CIA does not acknowledge error gracefully, and often does not forewarn policymakers of revised views before the information is published.

A policymaker who has made decisions or developed a negotiating strategy based on one assessment, only to see it change or to find that it was wrong, will not think fondly of the CIA or soon wish again to proceed on its assurances or assessments.

The CIA's relationship with Congress also is a special problem for policymakers, and it profoundly influences the agency's role. Virtually all CIA assessments go to the two congressional intelligence committees. Most go also to the appropriations, foreign relations and armed services committees. Eight congressional committees get the CIA's daily national intelligence report. In 1986 the CIA sent some 5,000 intelligence reports to Congress and gave many hundreds of briefings.

All this is new, having developed over the last decade or so. As a result, many senators and representatives are often as well informed about the CIA's information and assessments on a given subject as concerned policymakers. Moreover, this intelligence is often used to criticize and challenge policy, to set one executive agency against another and to expose disagreements within the administration.

Imagine the reaction of the Ford administration in the mid-1970s when it went to Congress to get additional military aid for Cambodia, only to be confronted by the legislators with a new intelligence assessment that the situation was hopeless. Imagine President Carter seeking a U.S. troop cut in South Korea only to find Congress aware of a new intelligence estimate that concluded that there were more North Korean divisions than was previously estimated. Imagine the reaction of a secretary of defense, seeking funds for a new weapon, only to be told on the Hill of intelligence that the Soviets could neutralize the weapon.

(Most specialists writing about the change in recent years in the balance of power between the executive and Congress on national security policy cite Watergate and Vietnam as primary causes. I believe there was a third principal factor: the obtaining, by Congress in the mid-1970s, of access to intelligence information essentially equal to that of the executive branch.)

This situation adds extraordinary stress to the relationship between the CIA and policy agencies. Policymakers' suspicions that the CIA uses intelligence to sabotage selected administration policies are often barely concealed. And more than a few members of Congress are willing to exploit this situation by their own selective use of intelligence that supports their views.

The oversight process has also given Congress -- especially the two intelligence committees -- far greater knowledge of and influence over the way the CIA and other intelligence agencies spend their money than anyone in the executive branch would dream of exercising, from expenditures in the billions of dollars to line items in the thousands. Congress has been immensely supportive and steadfast over the past 10 years in providing the resources to rebuild American intelligence. But I suspect it causes policymakers considerable heartburn to know that Congress may actually have more influence today over the CIA's priorities and its allocation of resources than the executive branch.

The result of these realities is that the CIA today finds itself in a remarkable position, involuntarily poised nearly equidistant between the executive and legislative branches. The administration knows that the CIA is in no position to withhold much information from Congress and is extremely sensitive to congressional demands; the Congress has enormous influence and information, yet remains suspicious and mistrustful. Such a central legislative role with respect to an intelligence service is unique in American history and in the world. And policymakers know it.

Now, let me turn to the CIA's role and relationship with the policymaker as seen from the agency's vantage point. The preoccupation of senior policymakers with current reporting is, from the CIA's perspective, a major problem. If, as I have been told, the average tenure of an assistant secretary in government is 21 months, a short-term focus is understandable; but it is nevertheless lamentable, and ultimately very costly to the country.

One of the CIA's greatest concerns over the years has been the unwillingness or inability of most policymakers to spend much time on longer-range issues -- looking ahead several steps -- or in helping to guide or direct the agency's long-term efforts. For many years the CIA has struggled, largely in vain, to get policy officials to devote time to intelligence issues other than those directly related to a crisis.

One reason Congress has assumed a larger role in these areas, in my view, is that policymakers in successive administrations have largely abdicated their intelligence-guidance responsibilities. For many years, trying to get senior policy principals to attend meetings to discuss longer-range intelligence requirements has been an exercise in frustration. Beyond the lack of help on requirements, the CIA gets little feedback on its longer-range work that might help improve its relevance to policymakers' needs.

In part because of their insufficient familiarity with intelligence work, too many policymakers arrive at their posts with an unrealistic expectation of what the CIA can do. When this expectation is disappointed, it often turns into skepticism that the agency can do much of anything. The apparent lack of CIA access to Soviet Politburo discussions, for example, leads some policymakers to question whether anything the agency says about Soviet intentions or politics has value. Similarly, the CIA has difficulty forecasting coups (which, of course, usually come as a surprise also to the targeted leader) or the results of difficult decisions not yet made by foreign governments.

It has been my experience over the years that the usual response of a policymaker to intelligence with which he disagrees or which he finds unpalatable is to ignore it; sometimes he will characterize it as incomplete, too narrowly focused or incompetent (and sometimes rightly so); and occasionally he will charge that it is "cooked" -- that it reflects a CIA bias.

In my 21 years in intelligence, I have never heard a policymaker (or anyone else for that matter) characterize as biased or cooked a CIA assessment with which he agreed. On subjects such as Vietnam, various aspects of Soviet policy and behavior, Angola, Lebanon and the effectiveness of various embargoes or sanctions, as well as on a number of other events and issues, the CIA's analysts have drawn conclusions that have dashed cold water on the hopes and efforts of policymakers. Sometimes the CIA analysts have been wrong, more often they have been right; but on problems both large and small the agency has not flinched from presenting its honest view.

It is important, also, to understand the difference between personal and institutional views. CIA directors are entitled to a personal view, and more than one has approved an estimate with which he disagreed -- and separately conveyed his own opinion to policymakers. For example, John McCone was alone in the intelligence community in predicting that the Soviets would put missiles in Cuba in 1962. Sometimes, as in this case, a director's personal view is insightful and correct; sometimes, it is not.

Imust add also at this point that the views of the CIA are sometimes overshadowed or set aside by policymakers or their staffs in favor of unanalyzed reporting or the assessments of other sources, or of interpretations of their own that support policy preferences or actions.

I believe this played an important role in the Iran arms-sales affair, as the CIA's formal assessments were misused or ignored by individuals pursuing their own agenda. No analytical expert in the CIA believed in 1985 that Iran was losing the war, and only one or two believed Iranian support for terrorism was waning. And no CIA publication asserted these things. But this was by no means the first instance of selective use of raw reporting or assessments. I have seen it done routinely in five administrations, on large issues such as Vietnam and the Soviet Union, and on lesser ones as well.

A final thought: to attempt to slant intelligence would not only transgress the single deepest ethical and cultural principle of the CIA, it would also be foolish -- it would presuppose a single point of view in an administration and would ignore the reality of congressional readership. Indeed, in my opinion, the sharing of intelligence with Congress -- where members of both parties, with a wide range of views and philosophy, all see the information -- is one of the surest guarantees of the CIA's independence and objectivity.

It is imperative that Americans know that the CIA's primary role remains the collection and analysis of information -- as opposed to the agency's important but secondary role in covert action -- and that these tasks are carried out with skill and integrity.

Robert Gates, a career intelligence officer, is deputy director of the CIA. This article is adapted from a forthcoming issue of Foreign Affairs.