To share or not to share: the compatibility of NATO’s nuclear sharing arrangements with the Non-Proliferation Treaty

Nuclear sharing at the NPT RevCon

On 27 April 2026, States will gather in New York for the start of the Eleventh Review Conference of the Non-Proliferation Treaty (NPT). The ‘RevCon’ comes at a particularly turbulent moment in time, marked by nuclear saber-rattling by Russia in the context of its war of aggression against Ukraine, a rapid expansion of China’s nuclear arsenal, and renewed ballistic missile launches by the DPRK, not to mention the controversy over Iran’s nuclear programme culminating in US-Israeli attacks against Iran’s nuclear facilities. Meanwhile, the US accused China of clandestinely conducting low-yield nuclear explosive tests, suggesting it may resume nuclear testing ‘on equal basis’, while France recently presented a new policy of “dissuasion avancée” (forward deterrence).

All these elements make for a tense geopolitical setting and a diplomatic quagmire. Whether the Vietnamese RevCon President will fare any better than his two predecessors in securing an agreed outcome document by the end of the conference in late May remains to be seen.

One topic that has risen to greater prominence in the run-up to the RevCon concerns the ‘nuclear sharing arrangements’ (NSAs) between the US and several European NATO Allies. Pro memorie, under the NSAs, the United States has deployed a limited number of B-61 nuclear weapons to certain locations in Europe (while no official figure is published, the total would appear to be but a small fraction of the number of non-strategic nuclear weapons deployed in Europe by Russia).

In recent years, a growing group of States from the Global South has denounced the NSAs as incompatible with the NPT. Critique of the NSAs has hardened, and taken a more ‘legalistic’ tone (see e.g. here).

In light hereof, this post examines the legal arguments raised against the NSAs, and submits that they are ultimately unconvincing. Inasmuch as the ‘ordinary meaning’ of the key NPT provisions offers no conclusive answer, we focus instead on other interpretative elements that would appear relevant under Articles 31(3)(b) and 32 of the Vienna Convention on the Law of Treaties (VCLT), namely the NPT’s travaux, as well as subsequent practice following its entry into force.

A closer look at Articles I-II NPT

Pursuant to Article I of the NPT, “[e]ach nuclear-weapon State Party to the Treaty undertakes not to transfer to any recipient whatsoever nuclear weapons … or control over such weapons … directly, or indirectly; and not in any way to assist, encourage, or induce any non-nuclear-weapon State to manufacture or otherwise acquire nuclear weapons …, or control over such weapons …”. Article II NPT conversely obliges non-nuclear-weapon States not to “receive” such control or assistance. The notions of “transfer” or “control” are not further defined in the treaty.

NATO Allies have consistently maintained that the NSAs do not entail a prohibited ‘transfer’ of (control over) a nuclear weapon in the sense of Articles I-II NPT since the weapons concerned remain under full US custody and control. Further, inasmuch as the NSAs have removed the incentive for other European countries to pursue a nuclear arsenal of their own, rather than undermining the ‘object and purpose’ of the NPT, they are deemed instead to promote non-proliferation. Beyond these interpretive arguments drawn from the ‘ordinary meaning’ of the text as well as the Treaty’s ‘object and purpose’, reference can also be made to the Treaty’s travaux and the context of its birthing process.

The genesis of Articles I and II NPT: a US-Soviet compromise

It is well-known that the US and the USSR acted as pen-holders in drafting Articles I and II. Within the Eighteen-Nation Disarmament Committee (ENDC), the two countries initially tabled radically different proposals. In particular, the US draft left room for forms of joint ownership and collective control over a strategic nuclear system involving both nuclear-weapon States and non-nuclear-weapon States – options that were excluded under the Soviet  draft.

Discussions took place at a time when the US and other NATO Allies sought to formalize nuclear consultation and planning in the NATO context – a process that would culminate in the creation of the Nuclear Planning Group (NPG) in December 1966. Initially, however, more ambitious proposals centered around the creation of a NATO ‘Multi-Lateral Nuclear Force’ (MLF) or an ‘Atlantic Nuclear Force’ (ANF), with scenarios including, for instance, the manning of jointly funded weapon systems by mixed-nationality crews from different NATO members. The creation of a NATO nuclear force was, however, strongly resisted by the Soviet Union, in particular out of fear that Germany could gain access to nuclear weapons. 

The impasse was ultimately resolved through behind-the-scenes discussions between the US and the USSR. Upon instruction of the respective foreign ministers, a joint working group was set up in October 1966 to prepare a draft treaty. Declassified transcripts of the meetings confirm that the USSR was made aware of the US position that it would not agree to any text which would rule out NATO NSAs.  

Following arduous negotiations, the two superpowers ultimately arrived at a compromise in the form of a joint draft simultaneously submitted by the US and the USSR, which abandoned references to ‘access’ or ‘emplacement’ – terms that had featured in the initial Soviet draft -, instead focusing on direct or indirect ‘control’ over nuclear weapons. In the words of US negotiator George Bunn (here at p. 39): 

“We won removal of Soviet language… that threatened to embarrass both NATO nuclear consultations and two-key arrangements. The Soviets won language banning transfers through a group of States.”

Before the joint draft was formally submitted to the ENDC in 1967, the US co-chair shared with his Soviet counterpart a set of Q&As which the US had previously presented to NATO Allies to inform them of the outcome of the US-Soviet negotiations. This document, which would later surface in the proceedings before the US Senate, asserted that the NPT “does not deal with allied consultations and planning on nuclear defense so long as no transfer of nuclear weapons or control over them results”, and “does not deal with arrangements for deployment of nuclear weapons within allied territory as long as these do not involve a transfer of nuclear weapons or control over them”.

In his authoritative study of the NPT negotiating history, Mohammed Shaker similarly concludes that the US-Soviet compromise resulted in “the dropping of the multilateral option by the United States. In return, the Soviet Union dropped its insistence on discontinuing the already existing nuclear-sharing arrangements” (see Vol. I at 233). Based on similar analyses of the NPT’s negotiating history, many scholars agree that the language of Articles I and II was not drafted with the intention of prohibiting nuclear-sharing arrangements (see e.g. Hayashi, at 475).

In the margin, an oft-overlooked point is that the Soviet Union may itself have had an interest in ‘containing’ the scope of the two provisions, as evidence suggests that the USSR also deployed nuclear weapons in eastern Europe in the context of the Warsaw Pact (see e.g. here, at 214). It follows that Moscow’s negotiating posture may have been driven less by a principled objection to all forms of nuclear deployment abroad, and more by a desire to constrain Western arrangements without foreclosing its own strategic flexibility.

Et alors?

One might of course object that any pact between the two Superpowers – especially one arrived at behind closed doors – as to the interpretation of Articles I and II NPT can hardly be taken to reflect the agreement of all States parties, and accordingly cannot be opposable to other States. Critics of the NSAs have indeed suggested that, during the NPT negotiations, third countries were left in the blind, both with respect to the nuclear sharing arrangements themselves, as well as with the US-Soviet agreement not to exclude such arrangements through Articles I-II NPT (see here, at 16-18). A closer look at the travaux nonetheless reveals a starkly different picture.

What lessons from the ENDC and UNGA First Committee records?

First, it is clear that the existence of the NSAs and their main features were well-known by the members of the ENDC, and by UN Members more broadly. On various occasions, the US explained that it retained “affirmative control” over its nuclear weapons deployed in Europe “at all times” (here, at 21); that consultation with NATO Allies did not entail giving non-nuclear weapon States information “on how to manufacture nuclear weapons” (here, at 20); and that control was ensured i.a. through “complex codes and electronic devices which prevent any unauthorized action” (here, at 14). And elsewhere: “[l]ike the Soviet Union’s Warsaw Pact allies, United States allies do receive training in dual-capability weapon systems… However, United States nuclear weapons remain firmly in United States hands” (here, at 14).

Second, as the US and USSR presented their drafts of what would later become Articles I and II, third States were fully aware of the importance of the issue. For instance, considering the different formulations in the initial drafts, Mexico referred to this as the real “crux proliferationis” (here, at 8), while Burma noted how the question of “nuclear-sharing arrangements within military alliances” constituted one of the main bones of contention in the negotiations (here, at 19, para. 52). Several States urged the two powers to work out a compromise between themselves and/or responded positively when bilateral negotiations produced a joint draft. India, for one, welcomed that “there is settlement now between the two alliances on the question of nuclear-sharing arrangements” (here, at 30, para. 79).

Third, and most important, various States confirmed that the NPT did not prohibit nuclear sharing. For example, Mexico stated that the draft treaty was “only designed to guarantee that the number of States having nuclear weapons in their territory or control shall not be increased, but without in any way limiting the capacity of the present nuclear Powers to install or emplace nuclear weapons in the territories of other States” (here, para. 79). Both in the ENDC and in the UNGA First Committee, some States expressed regret that the NPT did not prohibit the deployment of nuclear weapons on the territories of non-nuclear weapon States, and did not prevent the training in the use of nuclear weapons of the armed personnel belonging to those non-nuclear weapon States (see e.g. India (here, para. 27 and here, para. 115); Algeria (here, para. 68); Uganda (here, para. 49), and; Mauritius (here, at 202) – for an isolated counter-example see Nigeria (here, para. 54)). Romania even – unsuccessfully – tabled a proposal of its own, not so much to ban nuclear sharing, but rather to introduce a system of verification through the UN Security Council (here, para. 13). Crucially, the expressions of regret did not stop these countries from ultimately signing and ratifying the NPT – in full knowledge that nuclear sharing would not be banned.

Subsequent practice: reading the leaves

Leaving aside the NPT’s travaux, what lessons can be gleaned from the subsequent practice in the application of the Treaty? Of course, recent PrepComs and RevCons reveal that  various States criticize nuclear sharing as contrary to the Treaty’s spirit (e.g. Brazil, South Africa) or as a plain violation (e.g. Brazil, Egypt, Iran). Earlier RevCons nonetheless tell a different story.

For starters, at the first RevCon in 1975, a group of five States introduced a draft resolution (see here) ‘inviting’ negotiations on a new treaty that would provide for the withdrawal of nuclear weapons from the territories of non-nuclear weapons States Parties. If this demarche surely reflects dissatisfaction with nuclear sharing practices, it implicitly confirms such practices were not taken to contravene existing obligations under the NPT. The proposal itself was never adopted. Au contraire, States Parties instead  adopted a Final Declaration confirming that “the obligations undertaken under Article I and II of the Treaty have been faithfully observed by all Parties” (here, Annex 1, at 2). 

Questions pertaining to the compatibility of NATO’s NSAs only first emerged at the 1995 RevCon. Somewhat ironically, it appears that the first to raise the matter was Mexico – nota bene one of the original ENDC members, which had recognized throughout the negotiations that the NPT did not prohibit nuclear sharing (see above). While around eight other countries followed suit, several tended to challenge the NSAs in rather ambiguous terms – Tanzania, for instance, noted, somewhat circularly, that “the deployment of nuclear warheads in the territory of non-nuclear-weapon States would represent a violation of Article I if it was interpreted as a transfer of nuclear weapons” (see here, para. 24 – our emphasis). In turn, multiple NATO Allies responded by asserting that NATO’s NSAs complied strictly with the NPT, with Germany, for instance, stressing that “at no Conference of the Parties had there ever been any complaint that those agreements contravened articles I and II of the Treaty” (see here, para. 20). What is more, rather than joining the nascent chorus of critics, Russia – co-author of the original treaty text – appears not to have expressed any view on the issue. It would take until 2015 – long after the removal of Soviet nuclear weapons from eastern Europe – for Russia to shift sides and openly challenge the NSAs (see here).

Ultimately, States Parties agreed at the 1995 RevCon to the indefinite extension of the NPT without reopening or modifying the wording of Articles I-II, or otherwise clarifying their interpretation, at a moment that would have been particularly opportune to articulate concerns regarding the interpretation of the NPT’s core non-proliferation provisions. 

The picture that emerges is that criticism of the NSAs at the 1995 RevCon can only be seen as “too little, too late” – and a similar observation necessarily holds true for the more explicit and widespread criticism in more recent years. Indeed, from the perspective of treaty interpretation, such criticism cannot offset the negotiating history of the NPT, NATO’s decades-long nuclear sharing practice, and the consistent opposition of NATO Allies against any reading of Articles I and II NPT departing from the original US-Soviet compromise.

Turning back the clock? 

This post has sought to engage with the criticism that NATO’s NSAs run afoul of the NPT by retracing the origins of the key NPT provisions, as well as their treatment at the early RevCons. Our analysis ultimately supports the conclusion of Boothby and von Heinegg (among others) that “NATO’s sharing arrangements for basing US nuclear gravity bombs in Europe are in compliance with the NPT” (here at para. 1.123).

This is not to say that all is rosy in the land of non-proliferation, of course. Even if the total number of US nuclear weapons stationed on European soil may be only a fraction of what it once was, worldwide verifiable elimination of nuclear weapons in accordance with Article VI NPT remains far off. Yet, rather than investing time and energy in a revisionist effort to turn back the clock on the interpretation of Articles I-II NPT, the RevCon’s momentum offers States an opportunity to explore meaningful progress on nuclear disarmament and risk-reduction.

*The present blog post in written in the authors’ personal capacity.