The NATO action against Serbia has prompted two related but separate debates: one, is it prudent? - is it the best, or least costly, manner to deal with the threat Serbia poses to the Kosovar population?

The second, is it legal? - has NATO the right to do this, under international law and prevailing norms of international conduct? Most of those who argue against it on one ground do so on the other - i.e., for the majority of critics it is imprudent and illegal, for the majority of supporters it is prudent and legal.

It may turn out to have been imprudent but legal, occasioning more suffering to the Kosovar population than inaction. For some supporters, on the other hand, the issue of legality is secondary to that of prudence. Something had to be done, whatever the legal conditions involved.

Opposition to the action on legal grounds rests on two straightforward arguments. Firstly, Serbia is a sovereign state. It has not attacked another sovereign state as Iraq did Kuwait in 1990. It is therefore entitled to respect for its sovereignty under the UN Charter, classically stated in Article 2.4.

Secondly, the UN Security Council, the supreme authority in matters of the use of force in international relations, has not authorised military action against Serbia, as it did against North Korea in 1950, against Iraq in 1991, and against Serbian actions in Bosnia in 1995.

Against this, supporters of the action make several counter-arguments. The sovereignty argument falls on two counts. One is specific to Kosovo: the three UN Security Council resolutions on Kosovo (1160, 1199, 1203) categorised Serbian actions in that province as a threat to international peace and security - i.e., as going beyond its sovereignty alone. Serbia was enjoined to take certain actions (withdrawing forces, collaborating with international peace monitors) which it has not taken. Serbia is, therefore, implicitly in violation of UN Security Council resolutions concerning international peace.

The second reason why the sovereignty argument falls is that the UN Charter itself allows for actions by member-states that override sovereignty, not only where international peace and security are threatened, but also where serious violations of humanitarian law occur.

We are liable, in the light of the dozens of humanitarian agreements reached after 1945, to focus on these subsequent agreements, concerning human rights, genocide, torture, treatment of civilians and the conduct of war generally. But the charter itself upholds general human rights principles - in the Preamble and in Article 13.

The charter, therefore, contains a contradiction between the rights of states to respect for their sovereignty and the rights of individuals and groups within states to international protection; it is one that has been there from the start. To this may be added Article 24 which enjoins on permanent members of the Security Council a "special responsibility" for the maintenance of international peace and security.

This right to take action, in a situation of grave violation of humanitarian law, is compounded by the right of NATO, as a regional security organisation, to take action, in its sphere of geographic responsibility, to preserve peace.

The charter does allow for regional organisations to play a security role and in recent years there has been greater stress on this, within the emerging model of "global governance": this model calls for devolution of responsibility, in economic, humanitarian and security issues from the centre, in New York, to regional bodies.

In Africa the Organisation of African Unity, in the former Soviet republics the CIS, in the Middle East the Arab League have all played some peacekeeping role. Indeed, Article 52 of the charter allows for action by regional bodies, provided such actions are consistent with UN aims.

But such an opportunity is itself contained by the requirement that the actions of regional authorities receive authorisation from the Security Council thus batting, in the Kosovo case, the ball back to whether actions are in accordance with Security Council resolutions.

At this point, however, the argument on law must leave, or least relate to, the argument on reality. Those in favour of the Kosovo action will claim that something had to be done: a plausible reading of the Security Council resolutions, and of the charter, is better than inaction. The call to get a further Security Council authorisation is a call for inaction since the Russians and the Chinese would veto it.

Some, such as the German Foreign Minister, Mr Joschka Fischer, have gone further. He has argued there are occasions where, even in the absence of a mandate, it is better to act than not to. A strong humanitarian case can be made for this. The ironic outcome here, and in the recent bombing of Iraq where no clear, or further, mandate was received, is that those arguing the most for an ethical stand (Fischer and Cook among them) are also those who are the first to dispense with full legal authorisation.

The reality argument can, of course, operate on the side of those opposed to action. For such critics the legal and moral arguments are a cover for the exercise of power by NATO states pursuing selfish agendas. At the more sober end of the argument, this includes defending NATO's credibility and that of other peace-enforcement operations, notably that in Iraq.

At the more extreme end, it includes pursuit of an anti-Serbian agenda, above all by Germany and Turkey, and even a tacit collusion with Milosevic to prolong the war until Serbia has expelled most Albanians from Kosovo. The critics reinforce this by pointing to the selectivity of Western and NATO moral indignation: if human rights violations of a serious kind entail intervention, it is said, then this should have led to operations elsewhere.

There has been much discussion in recent years of the growth of a new international humanitarian consensus, be this in regard to the right of intervention or concerning human rights, through such bodies as war crimes tribunals and the establishment of an International Criminal Court.

There is some substance to this, as the detention of Pinochet and the establishment of the ICC indicate. But we are a long way from a situation in which such agreements can operate irrespective of the interests of states. There will long be an overlap between self-interest and humanitarian action, leading, as a consequence, to selectivity.

Self-interest, as construed by politicians and, not least, their electorates, will dictate the limits of those interventions and other humanitarian actions that do take place. This will also mean that selectivity will continue to operate, justified on grounds of practicality and by the lame but not invalid argument that it is better to try to do some good than none at all.

The question Kosovo poses is not, therefore, whether a legal case can be made for the NATO action. It can. Nor is it whether an element of self-interest, and selectivity, operates. It does. It is rather whether, in the circumstances, this was the most prudent policy to pursue, in the light of what Serbia's response might be, and in the light of the constraints which, in the name of the very self-interest that the critics denounce, operate on the NATO operation.

Fred Halliday is professor of international relations at the London School of Economics.