The dilemma of humanitarian intervention has been overtaken since the terrorist attacks of 11 September 2001 with other preoccupations, but it has not been resolved and it has not gone away. When, if ever, is it appropriate for states, individually or collectively, to take coercive action, and in particular military action, against another state - not for the purpose of self defence, and not in order to address some larger threat to international peace and security as traditionally understood, but for the purpose of protecting people at risk within that state?

The issue was the subject of countless debates through the 1990s, not least for NATO. The main cases - ones both when intervention took place, and when it did not - are etched in our memory. None of them was well or confidently handled: the debacle of international intervention in Somalia in 1993; the pathetically inadequate response to genocide in Rwanda in 1994; the failure of the UN presence to prevent murderous ethnic cleansing in Srebrenica in Bosnia and Herzegovina in 1995; and then NATO's intervention, without Security Council approval, in Kosovo in 1999.

Every one of the big cases generated major international controversy, but usually too late to be useful, and never enough to settle the issues of principle once and for all, including the role and responsibility of the United Nations, and the nature and limits of state sovereignty. UN Secretary-General Kofi Annan challenged the General Assembly in 1999, and again in 2000, to find a way through these dilemmas, posing the issue in the starkest of terms: "If humanitarian intervention is, indeed, an unacceptable assault on sovereignty, how should we respond to a Rwanda, to a Srebrenica - to gross and systematic violations of human rights that affect every precept of our common humanity?"

His question, however, went unanswered. Advocates of intervention on humanitarian or human rights grounds, and anxious defenders of state sovereignty, dug themselves deeper and deeper into opposing trenches from which they have still not emerged. The new century began with intense disagreement persisting as to whether there is a right of intervention, how and when it should be exercised, and under whose authority.

Since the terrorist attacks of 11 September 2001, attention has shifted to other dilemmas: how to capture and punish terrorists, how to mount sustainable defences against them and the states who support them, and whether it is ever permissible to mount pre-emptive attacks against countries believed to be irresponsibly acquiring weapons of mass destruction. But echoes of the older debate persist even in this new context. US President George W. Bush - and to some extent UK Prime Minister Tony Blair - have repeatedly sought to bolster the case for military action against Saddam Hussein by citing the Iraqi regime's "brutal suppression of its own people". And in the case of Chechnya, anxieties persist as to whether Russia's right to wage internal mayhem in the name of confronting terrorism should really be allowed to continue unrestricted and unchallenged.

Elsewhere, it is only a matter of time before reports emerge again from somewhere in the world of massacres or mass starvation, or rape or ethnic cleansing, occurring or apprehended. The possibility of Zimbabwe embarking on a deliberate policy of not just suppressing but starving is a chilling current case in point. And then the question will arise all over again in the Security Council and in capitals and in the media - what do we do? This time round we must have the answers.

It was to try to provide answers that Canada established in September 2000 the International Commission on Intervention and State Sovereignty, which I co-chaired along with the Algerian diplomat and UN Special Adviser Mohamed Sahnoun. Our colleagues were highly experienced, high-profile and globally representative analysts and practitioners: Gisele Cote-Harper, Lee Hamilton, Michael Ignatieff, Vladimir Lukin, Klaus Naumann, Cyril Ramaphosa, Fidel Ramos, Cornelio Sommaruga, Eduardo Stein and Ramesh Thakur. We consulted comprehensively, meeting in Africa and Asia as well as Europe and North America, and holding roundtables and other consultations in China, Latin America, the Middle East, and Russia.

The Commission recognised from the outset that if its report was to be useful it had to be not only intellectually satisfying, but practical and politically savvy: capable of mobilising support from both North and South, and actually guiding and motivating action. To bridge the gulf between state attitudes it had to be innovative, not just restating the familiar but unhelpful academic refrain that sometimes hard choices had to be made between what was "legal" and what was "legitimate".

The course we chose was to turn the debate on its head, and to recharacterise it not as an argument about the "right to intervene" but rather about the "responsibility to protect". Casting the issue in this way has four big advantages. It looks at the issues from the perspective of those seeking or needing support, rather than those who may be considering intervention. The spotlight is back where it should always be: on the duty to protect communities from mass killing, women from systematic rape and children from starvation. It implies that the primary responsibility rests with the state concerned, and that it is only if the state is unable or unwilling to fulfil the responsibility to protect, or is itself the perpetrator, that it becomes the responsibility of the international community to act in its place. To "protect" implies more than to "intervene". It embraces not just a responsibility to react, but to prevent and rebuild as well. Both of these dimensions have been much neglected in the traditional humanitarian intervention debate, and bringing them back to centre stage, to rank in priority alongside reaction, makes reaction itself - in appropriate cases - more palatable. Above all, new language helps de-prickle the policy debate: the actors have to change their lines, and think afresh about what the real issues are. The language of humanitarian intervention - which itself has been so divisive, offending those who have hated any association of the word "humanitarian" with military activity - is no longer the language of the debate. Consensus becomes easier to find.

The starting point in justifying this conceptual shift is the concept of state sovereignty itself. We argued that its essence should now be seen not as control but as responsibility. A large and growing gap has been developing between the codified best practice of international behaviour as articulated in the UN Charter, whose explicit language emphasises the respect owed to state sovereignty in its traditional Westphalian sense, and actual state practice as it has evolved in the 56 years since the Charter was signed. The new focus on human rights and, more recently, on human security, emphasises the limits of sovereignty. The Commission was intrigued to find, in our worldwide travels, just how much that gap was acknowledged. The defence of state sovereignty, by even its strongest supporters, did not include any claim of the unlimited power of a state to do what it wants to its own people.

We did not argue that there is now a sufficiently strong basis in principle and practice to claim the existence of a formal new principle of customary international law. But we did argue that the "responsibility to protect" is an emerging international norm, or guiding principle of behaviour for the international community of states, which may well become customary international law if further consolidated in state and intergovernmental practice.

Whatever else it encompasses, the responsibility to protect implies above all else a responsibility to react to situations of compelling need for human protection. When preventive measures fail to resolve or contain the situation, and when a state is unable or unwilling to redress the situation, then interventionary measures by other members of the broader community of states may be required. These coercive measures may include political, economic or judicial measures, and in extreme cases - but only extreme cases - they may also include military action.

But what is an extreme case? Where should we draw the line in determining when military intervention is, prima facie, defensible? What other conditions or restraints, if any, should apply in determining whether and how that intervention should proceed? And, most difficult of all, who makes all these decisions: who should have the ultimate authority to determine whether an intrusion into a sovereign state, involving the use of deadly force on a potentially massive scale, should actually go ahead? These questions have generated an enormous literature, and much competing terminology, but on the core issues there is a great deal of common ground. All the relevant decision-making criteria seemed to the Commission to be subsumed under the following six headings, involving a threshold criterion, four precautionary criteria and an authority criterion.

Just cause threshold

For military intervention for human protection purposes to be warranted, there must be serious and irreparable harm occurring to human beings, or imminently likely to occur, of the following kind: large-scale loss of life, actual or apprehended, with genocidal intent or not, which is the product either of deliberate state action, or state neglect or inability to act, or a failed state situation; or large-scale ethnic cleansing, actual or apprehended, whether carried out by killing, forced expulsion, acts of terror or rape.

The threshold needs to be set high and tight, for both conceptual reasons (military intervention must be very exceptional) and practical political ones (if intervention is to happen when it is most necessary, it cannot be called upon too often). Two situations only are identified as legitimate triggers. No attempt is made to quantify what is "large scale", but it is made clear that military action can be legitimate as an anticipatory measure in response to clear evidence of likely large-scale killing or ethnic cleansing. Without this possibility of anticipatory action, the international community would be placed in the morally untenable position of being required to wait until genocide begins, before being able to take action to stop it.

The threshold criteria articulated are wide enough to cover not only the deliberate perpetration of horrors such as occurred, or were anticipated, in Bosnia and Herzegovina, Kosovo and Rwanda, but situations as well of state collapse and the resultant exposure of the population to mass starvation and/or civil war (as in Somalia). Also potentially covered would be overwhelming natural or environmental catastrophes, which are not in themselves man-made, but where the state concerned is either unwilling or unable to cope, or call for assistance, and significant loss of life is occurring or threatened.

What are not covered by the "just cause" threshold criteria as set out here are situations of human rights violations falling short of outright killing or ethnic cleansing (such as systematic racial discrimination or political oppression), the overthrow of democratically elected governments and the rescue by a state of its own nationals on foreign territory. Although eminently deserving of external action of various kinds - including in appropriate cases political, economic or military sanctions - these are not cases which would seem to justify military action for human protection purposes.

Right intention

The primary purpose of the intervention, whatever other motives intervening states may have, must be to halt or avert human suffering.

There are a number of ways of helping ensure this. One is to have military intervention always take place on a collective or multilateral rather than single-country basis. Another is to look to whether, and to what extent, the intervention is actually supported by the people for whose benefit the intervention is intended. Yet another is to look to whether, and to what extent, the opinion of other countries in the region has been taken into account and is supportive.

The absence of any narrow self-interest may be an ideal, but it is not likely always to be a reality. Mixed motives, in international relations as everywhere else, are a fact of life. Moreover, the budgetary cost and risk to personnel involved in any military action may in fact make it politically imperative for the intervening state to be able to claim some degree of self-interest in the intervention, however altruistic its primary motive.

Last resort

Military intervention can only be justified when every non-military option for the prevention or peaceful resolution of the crisis has been explored, with reasonable grounds for believing lesser measures would not have succeeded.

The responsibility to react - with military coercion - can only be justified when the responsibility to prevent has been fully discharged. This does not necessarily mean that every such option must literally have been tried and failed: often there will simply not be the time for that process to work itself out. But it does mean that there must be reasonable grounds for believing that, in all the circumstances, if the measure had been attempted it would not have succeeded.

Proportional means

The scale, duration and intensity of the planned military intervention should be the minimum necessary to secure the defined human protection objective.

The action taken has to be commensurate in scale with its stated purpose, and in line with the magnitude of the original provocation. The effect on the political system of the country targeted should be limited to what is strictly necessary to accomplish the purpose of the intervention.

Reasonable prospects

There must be a reasonable chance of success in halting or averting the suffering which has justified the intervention, with the consequences of action not likely to be worse than the consequences of inaction.

Military action can only be justified if it stands a reasonable chance of success, and will not risk triggering a greater conflagration. Application of this precautionary principle would on purely utilitarian grounds be likely to preclude military action against any one of the five permanent members of the Security Council, even with all other conditions for intervention met: it is difficult to imagine a major conflict being avoided, or success in the original objective being achieved. The same is true of other major powers.

This raises the familiar question of double standards. Here, the only answer is that the reality that it may not be possible to intervene in every case where there is justification to do so, is no reason never to intervene.

Right authority

There is no better or more appropriate body than the United Nations Security Council to authorise military intervention for human protection purposes. The task is not to find alternatives to the Security Council as a source of authority, but to make the Security Council work better than it has.

When it comes to authorising military intervention for human protection purposes, the argument is compelling that the United Nations, and in particular the Security Council, should be the first port of call. The difficult question - starkly raised by Kosovo - is whether it should be the last.

The issue of principle here was in the Commission's view unarguable. The United Nations is unquestionably the principal institution for building, consolidating and using the authority of the international community. Those who challenge or evade the authority of the United Nations as the sole legitimate guardian of international peace and security in specific instances run the risk of eroding its authority in general and also undermining the principle of a world order based on international law and universal norms.

If the Security Council is for any reason unable or unwilling to act in a case crying out for intervention, there are on this view only two institutional solutions available. One is consideration of the matter by the General Assembly in Emergency Special Session under the "Uniting for Peace" procedure (used as the basis for operations in Korea in 1950, Egypt in 1956 and the Congo in 1960), which may well in fact have delivered, and speedily, a majority recommendation for action in the Rwanda, and especially Kosovo, cases. The other is action by regional or sub-regional organisations under Chapter VIII of the Charter within their area of jurisdiction, subject to their seeking subsequent authorisation from the Security Council (as happened with the West African interventions in Liberia in the early 1990s and Sierra Leone in 1997).

Interventions by ad hoc coalitions (or, even more, individual states) acting without the approval of the Security Council, or the General Assembly, or a regional or sub-regional grouping of which the target state is a member, do not - it would be an understatement to say - find wide international favour. There are many reasons to be dissatisfied with the role that the Security Council has played so far - its generally uneven performance, its unrepresentative membership, and its inherent institutional double standards with the Permanent Five veto power. But the political reality is that if international consensus is ever to be reached about when, where, how and through whom military intervention should happen, it is very clear that the central role of the Security Council will have to be at the heart of that consensus.

But what if the Security Council fails to discharge its own responsibility to protect in a conscience-shocking situation crying out for action, as was the case with Kosovo? A real question arises as to which of two evils is the worse: the damage to international order if the Security Council is bypassed, or the damage to that order if human beings are slaughtered while the Security Council stands by? The Commission's response to this dilemma was to articulate two important, essentially political, messages.

The first message is that if the Security Council fails to act, other states may act - and get it wrong. Such interventions, without the discipline and constraints of UN authorisation, may not be conducted for the right reasons or with the right commitment to the necessary precautionary principles. The second message is that if the Security Council fails to act, other states may act - and get it right. The ad hoc coalition or individual state may fully observe and respect all the necessary threshold and precautionary criteria, intervene successfully, and be seen to have done so by world public opinion - with this then likely to have enduringly serious consequences for the stature and credibility of the United Nations itself. That is pretty much what happened with the NATO intervention in Kosovo, and the United Nations cannot afford to drop the ball too many times on that scale.

The bottom line of the Commission's report is that when the next case of threatened mass killing or ethnic cleansing comes along, as it surely will, it must be dealt with expeditiously, and in a systematic, thoughtful and above all principled way. The erratic indifference of the 1990s must not be repeated. A good place to start in ensuring this would be agreement by the Security Council, at least informally, systematically to apply the principles here set out in any such case. So too would be a declaratory UN General Assembly resolution giving weight to those principles and the whole idea of the "responsibility to protect" as an emerging international norm.

We cannot be content with reports and declarations. If we believe that all human beings are equally entitled to be protected from acts that shock the conscience of us all, then we must match rhetoric with reality, and principle with practice. We must, as an international community, be prepared to act. There must be no more Rwandas, and no more Srebrenicas.

*The findings of the International Commission on Intervention and State Sovereignty are contained in the report The Responsibility to Protect that is available at www.iciss-ciise.gc.ca