When did the Armed Attack against Ukraine become ‘Imminent’?

Written by

When did Russia’s armed attack on Ukraine begin? And, before it began, when did it become imminent, as that term is commonly understood in the international law on the use of force? In this post I will offer some thoughts on these two questions, not because they are directly relevant to the situation in Ukraine – they are not, because Ukraine’s defensive actions against Russia were not pre-emptive – but because, I submit, Ukraine as a case study can teach us some useful lessons on the workability of the notion of imminence. And that notion may be central in evaluating future conflicts, some of which may be at least as dangerous as Russia’s assault on Ukraine. If, say, the United States or Israel ever choose to intervene militarily against Iran or North Korea, the justification they would offer would almost certainly revolve around a necessity to stop an imminent armed attack.

Some background

Whether Article 51 of the UN Charter permits recourse to self-defence against armed attacks that are yet to occur has been a perennial question of the jus ad bellum. Doctrine has discussed preventive, pre-emptive, anticipatory and interceptive self-defence endlessly. It has done so in conventional interstate contexts, and in those involving attacks by (terrorist) non-state actors, and regarding novel challenges posed by cyber. While the views of scholars (and states) on these issues are varied, they can still usefully be put in several big groups in light of the formal and policy arguments they make.

On one end of the spectrum is the restrictivist position: the text of Article 51 is clear and speaks of the right to self-defence arising only if an armed attack ‘occurs’; there can be no self-defence against a future attack. On the other (in every sense extreme) end of the spectrum are broad conceptions of preventive self-defence against future threats, especially those involving terrorism and weapons of mass destruction. This is a position most commonly associated with the George W. Bush administration in the United States, but its main problem is simply that the idea that ‘defensive’ force can be used to prevent highly speculative future threats cannot be made compatible with a legal system that comprehensively prohibits the unilateral use of force in international relations.

Between utopia and apology and attempting to occupy the middle ground are those who argue that self-defence is possible against a future armed attack, but only very exceptionally, if the attack is imminent. Virtually all states and scholars that endorse some variant of self-defence in response to a future attack thus rely on imminence as a limiting criterion – it would be unreasonable for a victim state to have to suffer the weight of the first blow before it could respond, but it would also be unreasonable to launch pre-emptive responses to any threat that is not genuinely imminent. Thus, for example, before the US/UK invasion of Iraq in 2003 the official legal advice of the UK Attorney-General disavowed any broad theory of preventive self-defence – like the one offered in the US Department of Justice memo providing legal cover for the war within the Bush administration – and expressly invoked imminence as a limiting criterion, concluding that on no reasonable view could Iraq be said to be imminently attacking the United States or its allies (para 3).

The big question here is of course what imminence actually means. Here again we have two big groups of positions.

On one view imminence is a temporal criterion – an imminent armed attack is one that is about to occur. On the other view an imminent attack is one that will occur in the ordinary course of events, and it is necessary to act now to deflect it. Imminence is here an aspect of the necessity of self-defence, and contains elements of causality and intention. Here’s how one proponent of such a theory, my good friend Mike Schmitt, puts it (p. 535):

the correct standard for evaluating a preemptive operation must be whether or not it occurred during the last possible window of opportunity in the face of an attack that was almost certainly going to occur. Restated, it is appropriate and legal to employ force preemptively when the potential victim must immediately act to defend itself in a meaningful way and the potential aggressor has irrevocably committed itself to attack.   

The key issue here is what we can mean by the aggressor having ‘irrevocably committed itself to attack,’ which I will come back to shortly.

Additionally, since 9/11 we have seen a concerted effort by some powerful Western states at stretching the notion of imminence (seen more as an aspect of the necessity of the response than as a temporal quality of the attack), especially in the context of attacks by terrorist non-state actors. The stretching has at times been so significant that it has become difficult to see what the real difference is, other than the packaging, between such approaches and the broad theories of prevention that are universally regarded as unacceptable in the modern international legal order. A pivotal part of that effort have been the so-called Bethlehem Principles, whose approach to imminence was adopted verbatim in the speeches by (for example) the British and Australian attorneys-general, invoking also Mike Schmitt’s idea of a last window of opportunity. (For detailed analysis, see this article by Dapo Akande and Thomas Lieflander, here at 8-10, and my prior posts here and here). But these expansive approaches to imminence cannot logically or legally be confined to attacks by terrorists; whatever imminence means, it must mean the same thing if the future armed attack is originating from a state, rather than from a non-state actor.  Recent high-profile invocations of self-defence against imminent attacks by states include (at least arguably so) the US assassination of the Iranian general Soleimani in January 2020, and the ongoing strikes by Israel against Iranian assets in Syria and Lebanon.

Occurrence, imminence and Ukraine

This brings us now to Ukraine. Obviously, Ukraine did not pre-emptively act against Russia’s attack (quite the reverse – it is Russia that seems to be making some kind of pre-emptive or preventive self-defence justification for using force). Even if it was 100% certain that Russia would launch its invasion on 24 February, Ukraine had no feasible way of deflecting this massive attack pre-emptively. And it was not in Ukraine’s strategic interest to fire the first shot for what would, at best, be minor tactical benefits, while giving Russia a pretext for the invasion. Ukraine’s situation is therefore different from those cases in which states actually relied on self-defence to respond before an attack came. But this is precisely why this is a good case to test out intuitions about imminence as a quality or characteristic of the armed attack, rather than as an aspect of the immediacy/necessity of the response to the attack.

In other words, at some point before Russia’s armed attack against Ukraine started occurring, it must have become imminent in some logical and legal sense, in accordance with those interpretations of Article 51 of the Charter that allow for anticipatory or pre-emptive self-defence before an attack occurs. The imminence of Russia’s attack is a feature of that attack regardless of the fact that Ukraine did not take advantage of the right it arguably had to respond before the attack landed. In short, the ‘imminence’ of the attack cannot depend on whether Ukraine chose to respond to it pre-emptively or not. Scholars (and states) who argue that an ‘imminent armed attack’ is a coherent concept therefore need to be able to answer (1) more straightforwardly, when Russia’s attack against Ukraine actually began, i.e. started occurring within the meaning of Article 51, and then (2) at what point in time before the moment of occurrence that attack became imminent, as opposed to merely potential.

On the first question one could argue that Russia’s armed attack against Ukraine began all the way back in 2014 with the annexation of Crimea and intervention in the Donbas, and that legally the full-scale invasion was simply a continuation of what came before it. That’s true. But, at least as I can judge state views on this, it also seems reasonably clear that the invasion in 2022 has been regarded as a game-changing separate event – a new use of force requiring its own separate justification – by all parties concerned. And for the purpose of understanding imminence we could simply postulate the prior uses of force away.

If so we could (I imagine) all agree that the armed attack against Ukraine began shortly after 6am Moscow/Kyiv time on 24 February, as Putin wrapped up his speech announcing the ‘special military operation’ and missiles started raining down on Ukrainian cities, with ground forces moving into the country while the UN Security Council was in session. I also imagine that we would all agree that the armed attack started occurring some minutes before the missiles exploded, at the moment they were actually fired against their targets (see discussion here). Some might also argue that the attack started occurring at an earlier point, e.g. when Putin gave the very final order for the invasion to proceed, which was presumably in the hours before (but we don’t know exactly when).

Let’s say we fix the moment of the occurrence of the attack at 6am Moscow/Kyiv time on 24 February. When, then, did the attack become imminent? On a strict temporal construction of imminence (i.e. an imminent armed attack is one that is about to occur), the attack could have been imminent in the hours before it began, or the day before it began, or maybe even a week or two before it began. But on a causal construction of imminence, the inquiry would be into determining that point in time at which ‘the potential aggressor has irrevocably committed itself to attack’, to quote from Mike Schmitt’s formulation above, which is as good as any.

Imminence, capabilities, intentions

This is then the key question – at what point exactly did Russia irrevocably commit itself to attacking Ukraine? What does this mean? And how can we know this?

Any assessment of imminence in this sense must be about two things: the attacker’s capability, and the attacker’s intentions. (Again, I am here trying to clearly separate the imminence of the attack in this sense from the necessity/immediacy of any pre-emptive response to the attack). The mere capability to attack cannot make an attack imminent without the corresponding intention – the United States and Russia have the capability to destroy each other (and the world) within hours using nuclear weapons, but this capability does not ipso facto mean that any such attack is imminent. Similarly, intention to attack is meaningless without the capability to pursue the attack (which, as the war in Ukraine teaches us, can nonetheless be far removed from the attacker’s capability to actually achieve their stated goals).

Assessing when Russia’s attack against Ukraine became imminent in the legal sense thus requires an inquiry, ex ante or ex post, into Russia’s capabilities and intentions. The former is of course much easier than the latter, since the accumulation and placement of military assets can be objectively observed, e.g. through satellite imaging. Russia’s military capabilities were substantial to begin with, and went through two period of sustained build-up – first in the spring of 2021, and then from November 2021 up to the assault in February 2022. Russia’s ability to quickly launch a massive attack on Ukraine simply became impossible to deny.

But the question of Russia’s intentions is more complicated. And we are indeed here talking about subjective intentions, because there is no other way of conceptualizing the imminence of an event that is the product of human action. Here, at least, we don’t have any real difficulty in anthropomorphising the abstract entity of the Russian state – the key decisions here were all made by a single individual, Vladimir Putin. In the autocratic regime he created, it is his intentions alone that really mattered, and it is he alone who (at some point) irrevocably committed to attacking Ukraine. What we don’t know – at least not yet, at least not on the information publicly available – is when exactly he made this decision. It could just as easily have been 23 February 2021 as 23 February 2022. And we don’t know that at least partly because Putin held his cards so closely to his chest that most of his closest advisors and military officers had no idea what his intentions were until the order to initiate the attack came.

Getting into Putin’s head, irrevocably

If the notion of irrevocability asks us to examine a state’s (its leader’s) intentions, we run into the basic problem that intentions and decisions can change. If interpreted literally, an irrevocable decision could be one that could not be changed even if the leader’s calculation behind it does change – but because modern technology allows for instant communication between the leader and those implementing his decisions, the temporal window for irrevocability would be very narrow indeed. Alternatively, irrevocability could be interpreted as a much less demanding standard – a reasonably firm decision by the leader to pursue the attack, unless the circumstances on the ground substantially change. But even on such a standard pinpointing the moment in time in which the decision was made may be difficult or impossible.

It would be good at this point to reflect on the (publicly disclosed) assessments of Western intelligence agencies in this regard. For weeks ahead of the invasion we could see a constant stream of statements by officials of various governments, especially the US, to the effect that the intelligence information at their disposal disclosed an imminent attack by Russia against Ukraine. But these statements were frequently mixed with messages from those states and others that a final decision to attack was yet to come.

Thus, for example, on 28 January President Zelensky called on Western states not to create a panic damaging to Ukraine’s economy through their many warnings, saying that ‘he did not see a greater threat now than during a similar massing of troops last spring.’ Despite the US warnings, the US Defence Secretary Lloyd Austin stated that ‘Conflict is not inevitable. There is still time and space for diplomacy,’ while the head of Germany’s foreign intelligence service said that he believed ‘that the decision to attack has not yet been made’ even if the capability was there.

Similarly, on 11 February the US National Security Adviser stated that Russia could invade Ukraine within days, but that ‘We have not seen anything come to us that says a final decision has been taken, [that] the go order has been given.’ An unnamed European diplomat insisted that it was not too late to deter Putin, saying that ‘There are further decision points along the line.’ And the head of the Norwegian intelligence service assessed that Russia had the capability to invade Ukraine, but that it was ‘first and foremost up to President Putin if he chooses to do so … It is difficult to say whether it is probable or not probable, because it is solely up to the Russian president to make that decision.’

A week or so later the language started firming up. On 17 February the US ambassador to the UN tweeted (in language that I imagine must have been lawyered) that ‘The evidence on the ground is that Russia is moving toward an imminent invasion.’ And then on 18 February President Biden implied that a Rubicon had been crossed by saying that ‘As of this moment, I’m convinced he’s made the decision [to invade] … We have reason to believe that.’ On 20 February US officials briefed reporters to explain that President Biden’s assessment was made on the basis of new intelligence in which they had high confidence:

The U.S. has intelligence that Russian commanders have received orders to proceed with an invasion of Ukraine, with commanders on the ground making specific plans for how they would maneuver in their sectors of the battlefield, a U.S. official told CBS News.  The orders don’t mean a invasion is a certainty, as Russian President Vladimir Putin could still change the orders if he changes his mind, the official said.

Note again the mixed messaging – the orders to proceed were given, but Putin could still change them if he changes his mind. The orders were given, but the invasion was not a certainty. US Secretary of State Anthony Blinken similarly added that ‘Everything we’re seeing tells us that the decision we believe President Putin has made to invade is moving forward’ but also that ‘Even if the die is cast, until it’s settled, until we know that the tanks are rolling, the planes are flying, and the aggression has fully begun, we’re going to do everything we can to prevent it but we’re prepared either way.’ Blinken justified not triggering sanctions at that moment, despite Zelensky’s urging, because ‘Once you trigger the sanctions, you lose the deterrent effect … As long as there is still even a minute’s worth of time in which we can deter and prevent a war, we’re going to try to use it.’ The idea here, clearly, is that the deterrent effect of possible (imminent?) sanctions might make Putin change his mind (which we now know it didn’t).


This was hardly a comprehensive analysis of all of the relevant statements made in the run-up to the war, but the direction of travel is I hope clear. And despite the pervasive use of the terminology of imminence, at virtually all times, even when they said the orders to proceed with the invasion were given, Western officials emphasized the continuing possibility that President Putin would change his mind, as he perhaps had done in the spring of 2021, or had not yet made up his mind firmly. Notably, the US and the UK did not simply make their assessments internally, but engaged in a sustained joint effort to share intelligence with allies, including Ukraine, and to disclose some of their assessments publicly. This operation has been hailed as a significant intelligence success, especially when compared to the Iraq fiasco. But even so up until the very last minute some allies remained sceptical, with the French for example continuing to believe that a full-scale invasion was unlikely (on the consequent resignation of the head of French military intelligence, see here).

I will make three interrelated points here. The first is that the intelligence exercise here was essentially about what one man – Vladimir Putin – was thinking. Not only is that exercise inherently difficult, it is made even more difficult by the nature of the man, as explained in this New York Times report:

Mr. Putin’s calculus, according to a U.S. official, is likely shifting as he weighs the changing costs of an invasion and he assesses what he could get from negotiations. Several officials note that Mr. Putin has a history of waiting until the last possible moment to make a decision, constantly re-evaluating his options.

Not surprisingly, American officials will not say how they know what Mr. Putin is thinking, anxious to preserve their current sources.

Knowing the intention of any autocratic leader is difficult, but Mr. Putin, who began his career as a K.G.B. officer, is a particular challenge. Because he avoids electronic devices, oftentimes bans note-takers, and tells his aides little, there is a limit to how much an intelligence agency can learn about his intentions and thinking.

Second, we should also be clear that, when it comes to the key question of predicting Putin’s intentions and actions, the undoubted success of Western intelligence in both making the prediction and the public communication of that prediction did not entail the public release of any concrete evidence. We have seen none. We have no idea what the factual basis on which the relevant agencies made their assessments actually was. None of those assessments, whether we chose to trust them or not, could be validated by outside observers until the moment of the attack came. And this ‘faux transparency’ as Henning Lahmann aptly describes it – the public communication of intelligence assessments without any communication of supporting evidence – is a real problem in a situation of epistemic crisis that plagues so many different societies. It is also a pervasive problem in international law, where legal claims are often dependent on a factual predicate that cannot objectively be verified.

Third, if the legal standard of imminence is that a state – its leader – has irrevocably committed to attacking another state, the basic problem that we may have on our hands is that this standard might be too demanding, while any conceivable alternatives are too loose and permissive. When, exactly, did Putin irrevocably commit to invading Ukraine? I’m not sure we have an idea of this even now. Certainly the officials of the various states making this assessment ex ante could not do so with confidence. It is entirely possible that Putin, with his ‘history of waiting until the last possible moment to make a decision, constantly re-evaluating his options,’ really didn’t irrevocably make up his mind right until the morning of 24 February. And if that’s true (I am not saying it is, but it could be), then how could an imminence standard be workable?

Let me conclude by saying that I shouldn’t be taken as arguing here, in a fully restrictivist vein, against any possibility of self-defence until an armed attack actually occurs. I am genuinely agnostic on the question. But the proponents of theories endorsing some version of anticipatory self-defence should reflect on whether the notion of imminence that does so much work here in delineating between the justified and the unjustified can actually do the work that is expected of it. And in doing so they could try answering a simple counter-factual question: had Ukraine launched a missile barrage against Russian airfields or fuel depots on (say) 17 February or 23 February, before waiting for the first blow to come, would that have been a justified case of self-defence against an imminent armed attack?

Print Friendly, PDF & Email

Leave a Comment

Your comment will be revised by the site if needed.


Christian Henderson says

April 20, 2022

Great post, thanks Marko. It is that phrase ‘irrevocably commit’ which seems to be key, and in that sense an irrevocable decision is one where the possibility of reversing it has been lost. When the big red button has been pushed, if you like, and not simply a personal commitment by one individual that military action should be taken at some point. It’s perhaps difficult then to draw any real distinction between the concept of anticipatory action, based on imminence (which still attracts some controversy), and interceptive action to prevent the effects of an attack from occurring that is already (and irrevocably) launched (which most would presumably accept).

Mary Ellen OConnell says

April 20, 2022

Dear Marko,

The concept of "imminence" is not workable, which is why it does not appear in the Charter. If we have learned anything from the Russian invasion, it is to support the Charter as written.


Mary Ellen

Marcin J. Menkes says

April 21, 2022

So here we have another contribution to the irrevocably debate: in a televised spectacle, President Putin accepts a report on the seizure of Mariupol but cancels (last part of) the order to attack Azovstal steel plant. In other words, objective irreversibility as a criterion for evaluating irrevocable commitment is problematic since even an ongoing attack can be halted.


Eugenio Carli says

April 21, 2022

Dear Dr Milanovic,

Thanks for this post, which like all the others you have written, is particularly thought-provoking.

As Mary Ellen O'Connell commented, I also think the concept of imminence is not workable, at least not in objective and straightforward terms to make it a usable legal criterion. Too many factors and variables come into play which make the creation of a clear norm on this sensitive subject impossible, irrespective of the appropriateness of the terms used to try to depict the concept.

This is all the more so considering the interests at stake. In this regard, I would propose the following additional (and provocative) question to be inserted after yours: "What if the missile barrage by Ukraine in self-defence would then prove to be unjustified, meaning that Russia was not going to launch any attack?" (both because Putin changed his mind or Ukraine misinterpreted the situation or whatever other reason).

Michael Misty says

April 24, 2022

Hi Marko,
Yes, Russia's armed attack did began in 2014 and continues to this day. The fact that Western states undoubtedly saw the full-scale invasion in February as a geopolitical turning point meriting a heightened response is obviously not to say that collective self-defence was not a legal option earlier.

If, in this thought experiment, there had been no earlier/ongoing attack/annexation from 2014 (and if Russia possessed no nuclear weapons) the 100,000+ troops/materiel amassed on Ukraine's border for 'exercises'(despite idling around) would eventually have been viewed as a non-imminent threat of a use of force (in the 2(4) sense), not triggering a right of self-defence but permitting non-forceful countermeasures. If diplomacy and non-forceful CMs then proved ineffective at de-escalating you would have had both a stronger argument for there being an imminent threat of armed attack and the necessity of a forceful response on that basis. Btw, re 'faux transparency', all those troop movements were very publicly available on twitter.

Even articulating imminence in terms of irrevocability is a bit problematic. A missile that has been launched by an aggressor could potentially be diverted mid-flight (shift-cold). That's why the footnotes in the Bethlehem principles have qualifications like reasonable certainty etc. Similarly, what is construed as the 'last possible window for action' has to account for any declining probability of success in averting the attack with each passing window. Actually domestic criminal law is quite instructive.

Yes, there is a sense in which imminence speaks to necessity and immediacy. My feeling is that decision makers are driven by the former but the requirement for sufficient certainty of the necessity of action precludes more remote forecasting in temporal terms as well.

But this temporal issue isn't a debate only playing out in national security circles. Is the threat of climate change imminent? Incidentally, does the gravity of that global threat alter your necessity/temporal analysis? Not all armed attacks are created equal.

So, yes Ukraine could presumably have identified a point of at which the threat was imminent and acted under anticipatory SD. But for those strikes to have been effective - for them not to have run into necessity issues of another sort - they would surely have needed to be directed at something - or someone (see the first of your 3 points) - more significant than a fuel depot or airstrip. Not a good idea.