Prolepsia forecastYes / no
Will Russian forces and the armed forces of a NATO member clash directly, causing deaths, before 31 December 2027?
As of 18 November 2026, Prolepsia puts the chance of yes at 10%.
10%chance of yes
How Prolepsia reasons
The report Prolepsia wrote with this forecast.
How forecasts are made · Track record
TL;DR
Probability of Yes: 9.9%. Probability of No: 90.1%. The expected outcome is no verified lethal clash through December 30, 2027; the main risk is an isolated strike or encounter, not a wider war.
Key evidence
The clearest historical precedent is al-Bab, Syria, in February 2017, where a mistaken Russian airstrike killed three Turkish soldiers. The post-Soviet record contains few clear matches. The 2020 Balyun strike killed Turkish troops, but attribution to Russian rather than Syrian aircraft remains disputed.
Military contact has increased on NATO’s eastern flank. Romania’s defense ministry recorded nine unauthorized entries and no drone shootdowns in 2025, versus 39 entries and six shootdowns through November 17, 2026. That shows more encounters are progressing to weapons use, not that they are becoming lethal. A dangerous Russian intercept of a crewed British aircraft illustrates a more consequential route.
NATO-member personnel are also exposed inside Ukraine. A U.S. inspector general report recorded 65–75 personnel in the forward advisory contingent in Kyiv as of September 30. That is an existing presence, not a combat deployment. A separate UK–French multinational force plan remains conditional on hostilities ending. Danish intelligence assessed a low but rising risk of a limited Russian attack on a bordering NATO member.
Uncertainties
The biggest unknown is whether regular forces come into closer contact. A durable ceasefire without a contested deployment would reduce exposure. A larger official deployment in Ukraine, or weapons use against a crewed aircraft or ship, would raise it. Expanded EU maritime enforcement adds another route, but the November 13 boarding of the Midag was a flag inspection, and no recent boarding was verified to have occurred under Russian military escort. Attribution is a separate uncertainty: even a fatal incident does not count unless credible evidence connects the death directly to armed force involving both militaries.
Read the full report
TL;DR
Probability of Yes: 9.9%. Probability of No: 90.1%. No verified lethal clash is the expected outcome through December 30, 2027, and the main risk is an isolated strike or encounter rather than a decision to start a wider war.
Context
As of November 18, 2026, Russia’s war in Ukraine was still active. Talks in Abu Dhabi produced no broader breakthrough. NATO had condemned airspace violations involving Poland and Romania and continued to strengthen its eastern flank. The U.S. intelligence assessment warned that continued fighting increases the risk of accidental or deliberate escalation.
The question covers 408 calendar days, from November 18, 2026, through December 30, 2027, inclusive, anywhere in the world. One death directly caused by a qualifying engagement is enough. The other side need not return fire, and Article 5 need not be invoked. An injury inflicted by the deadline can count if the resulting death is confirmed by January 31, 2028. Nonfatal equipment losses and Ukrainian attacks using NATO-supplied weapons do not count on their own.
Evidence
The post-Soviet record gives a low historical anchor. The clearest match is al-Bab, Syria, on February 9, 2017. A Russian airstrike mistakenly hit Turkish troops and killed three. That case establishes why the forecast is not just about the risk of war: a targeting mistake can satisfy the rule even when neither government wants a fight. There is one clear match in the roughly 35 years of post-Soviet history reviewed. That is a sparse record, not a complete census, and it includes long stretches with much less direct military contact than today.
The next cases show why attribution matters. At Balyun in February 2020, Türkiye reported 34 soldiers killed and described Russian and Syrian aircraft operating together. Russia denied its aircraft carried out the strike, and the public evidence does not establish which force delivered the fatal weapon. In the 2015 Turkish shootdown of a Russian Su-24, the pilot survived ejection and was reportedly killed by rebel fire, not by the Turkish missile. At Khasham in 2018, lethal U.S. fire is established, but regular Russian military participation is not. A June 2000 killing surfaced in the militarized disputes record, but the Russian serviceman belonged to a separate border service. The evidence does not establish that he was operating as part of the regular armed forces.
Today’s eastern-flank exposure is higher than the long-run record suggests. Romania’s defense ministry recorded 185 nearby attack situations, 131 scramble situations, and 58 unauthorized entries from the start of the full-scale invasion through November 17, 2026. It recorded no drone shootdowns in 2022, 2023, 2024, or 2025. In 2025 it logged 28 nearby attack situations, 21 scramble situations, and nine unauthorized entries. In 2026 through November 17, it logged 118, 96, and 39, respectively, along with six drone shootdowns. Two of those shootdowns took place over Romanian territorial waters on October 17. The counts show that some encounters now progress from detection to weapons use. They do not provide a rate at which an encounter kills someone or meets the attribution rule.
Crewed encounters give a more direct picture of how a fatality could occur. A Russian fighter released a missile near a British surveillance aircraft in September 2022 without killing anyone. The UK reported another dangerous intercept over the Black Sea in April 2026. In September 2026, Danish Defence said a Russian frigate fired flares toward a Danish military helicopter. Russia disputed the account, and no one was killed. These encounters show how close forces can come without the contact becoming lethal. The Ukraine-war record reviewed contained no confirmed qualifying fatality as of the forecast date. That supports a low forecast, but it does not rule out an unreported incident.
Official personnel in Ukraine create a separate route. A U.S. inspector general report recorded 65–75 personnel in the forward advisory contingent in Kyiv as of September 30, against 93 authorized. France also confirmed a training and technical-support presence. These are real deployments, not evidence of a frontline combat mission. The larger UK–French multinational force plan is separate and conditional on hostilities ending. A strike that directly kills serving allied personnel in an official military capacity can qualify even if they are not there to fight.
Two weaker signals add to the risk. Denmark’s September 24 intelligence assessment identified a low but rising risk of a limited Russian military attack on a bordering NATO member even while the Ukraine war continues. That is not a prediction of a lethal outcome. At sea, the EU’s October 9 mandate expansion creates more scope for enforcement encounters. But the November 13 boarding of the Midag was a flag inspection, and the maritime review found no recent boarding under verified Russian military escort. Deadly incidents involving Black Sea merchant shipping show danger in the theater, not by themselves a clash between the two militaries.
What's non-obvious
More interceptions do not imply a proportional rise in lethal clashes. An uncrewed target can be destroyed without a human casualty. Even if falling debris kills someone, the death counts only if the evidence ties it directly to an armed engagement involving both militaries. The Romanian figures are therefore stronger evidence of more weapons contact than of more lethal contact. A single encounter involving a crewed aircraft can matter more for this question than several drone shootdowns.
The same distinction applies to Ukraine. An advisory contingent already in Kyiv creates exposure today; a much larger force that has only been announced does not. A ceasefire would reduce many opportunities for contact, but it would not necessarily reduce every route to a clash. A durable truce with no contested deployment would lower risk. A truce followed by a larger official deployment and then a return to fighting could put more regular forces in direct contact than they are now.
Uncertainties
The largest uncertainty is how often military contact progresses from monitoring or interception to force that can kill someone. The historical cases are too few to pin that down, and the modern incident counts mix alerts, incursions, and shootdowns. A sustained change in how either side treats crewed aircraft or ships would be more informative than a rise in raw scramble counts.
The scale, duration, and role of allied personnel in Ukraine are also uncertain. The September advisory count is a snapshot, not a forecast of future staffing. An implemented ceasefire, a confirmed expansion of the official military presence, or a verified encounter between an enforcement vessel and a Russian military escort would change the assessment more than another announcement or threat.
Finally, a fatal event may be easier to establish than the identity of the force that caused it. Balyun shows the gap: the deaths were documented, but the fatal strike was not conclusively attributed to Russian forces. Incident records and independent corroboration linking a weapon, the forces involved, and a death would narrow that gap. An allegation alone is not enough.
Forecast history
| Date | Forecast |
|---|---|
| 10% chance of yes |
Sources Prolepsia read
- US envoys go to Kremlin for talks with Putin on Ukraine – ThePrint – ReutersFeedtheprint.in
- Russian airstrike kills 3 Turkish soldiers in Syriawww.militarytimes.com
- S/PV.8738digitallibrary.un.org
- Turkey's Downing of a Russian Jetwww.csis.org
- Khasham in 2018www.oversight.gov
- Militarized Interstate Disputes (v5.0) – The Correlates of War Projectcorrelatesofwar.org
- Ministerul Apararii Nationalewww.mapn.ro
- Commons Chamber - Hansard - UK Parliamenthansard.parliament.uk
- RAF aircraft dangerously intercepted by Russian jets over Black Sea - GOV.UKwww.gov.uk
- Danish Defencewww.forsvaret.dk
The question
This question asks whether, between November 18, 2026, and December 30, 2027, inclusive, Russian armed forces and the armed forces of at least one NATO member state will engage in direct armed hostilities that result in at least one human death attributable to the engagement. The incident may occur anywhere in the world, including NATO territory, Russian territory, Ukraine, international waters, or international airspace. It may involve ground troops, naval vessels, military aircraft, drones, missiles, or other weapons systems. The confrontation need not escalate into a declared war or trigger NATO's Article 5 collective-defense provisions. Both deliberate attacks and unintended armed engagements, such as accidental exchanges of fire, qualify if they involve the military forces of both sides and result in at least one death. As of November 18, 2026, relations between NATO and Russia remain highly strained because of Russia's ongoing war against Ukraine, military activity along NATO's eastern flank, and concerns about escalation. In August 2026, NATO condemned Russian airspace violations involving Poland and Romania and announced continued efforts to strengthen its eastern defenses. NATO has also expanded its military presence along its eastern flank. The US intelligence community's 2026 Annual Threat Assessment warned that the continuation of the war in Ukraine increases the possibility of accidental or deliberate escalation into direct conflict between Russian and NATO forces. However, direct military confrontation is not inevitable, and NATO members continue to emphasize deterrence and defense. The forecast concerns an actual lethal armed engagement rather than diplomatic confrontation, military threats, airspace violations without lethal combat, or indirect participation in the Russia-Ukraine war. Background sources include NATO's August 12, 2026 statement on airspace violations (https://www.nato.int/en/news-and-events/articles/news/2026/08/12/nato-allies-condemn-recent-airspace-violations-and-incidents), NATO's overview of its eastern-flank defenses (https://www.nato.int/en/what-we-do/deterrence-and-defence/strengthening-natos-eastern-flank), and the US intelligence community's 2026 Annual Threat Assessment (https://archive.dni.gov/files/ODNI/documents/assessments/ATA-2026-Unclassified-Report.pdf).
How it resolves
The question resolves YES if credible evidence establishes that, from November 18, 2026, through December 30, 2027, inclusive, at least one direct armed engagement occurred between Russian military forces and the military forces of a NATO member state, and that the engagement caused at least one human death. A qualifying engagement must involve the use of physical military force by one side against the other, such as gunfire, artillery, missiles, aerial attacks, naval weapons, or armed drones. The other side need not return fire: a unilateral Russian military attack on NATO military personnel, or a NATO member's military attack on Russian military personnel, qualifies if it causes a death and meets the attribution requirements. Deaths may occur among Russian personnel, NATO member military personnel, or civilians directly killed by the engagement. The fatality must be directly attributable to the qualifying armed engagement, rather than to an unrelated accident or indirect consequence. The question resolves NO if no such qualifying engagement occurs within the specified period, or if armed incidents occur but none produces a verified death. The question also resolves NO if deaths result solely from cyberattacks, sabotage by intelligence operatives or nonmilitary proxies, attacks by Ukrainian forces using NATO-supplied weapons, or other indirect actions without direct participation by the regular armed forces of Russia and a NATO member state. Military personnel acting under NATO command or under their own national command qualify equally. The principal evidence will be official statements from NATO, the Russian Ministry of Defence, and the defense ministries or governments of the NATO member states involved, supplemented by independent reporting from Reuters, Associated Press, BBC, and other established international news organizations. A formal acknowledgment by both governments is not necessary. If a government disputes responsibility, the incident may still qualify when sufficiently detailed and independently corroborated evidence establishes the identity of the forces involved, the use of armed force, and at least one resulting death. Reporting must establish these elements with high confidence; unverified allegations or speculative attribution are insufficient. The outcome will be assessed using information available through January 31, 2028, allowing time for delayed reporting and investigations. If no qualifying event can be substantiated by that date, the question resolves NO, unless authoritative evidence establishes that a potentially qualifying incident occurred but the essential facts remain genuinely unresolved, in which case the question may be annulled rather than forcing an unsupported determination. Events occurring after December 30, 2027, do not qualify, even if they are part of a confrontation that began earlier.
Fine print
The deadline is interpreted literally as 'before December 31, 2027,' meaning the final qualifying date is December 30, 2027, at 23:59:59 UTC. Incidents that begin before the deadline but produce their first attributable death after the deadline do not qualify, unless the fatal injury was inflicted before the deadline and the victim subsequently died; in that case, the incident qualifies if the death is confirmed by the resolution date. At least one death is sufficient; there is no minimum number of combatants involved and no requirement for sustained fighting. Accidental friendly-fire incidents do not qualify unless they involve opposing Russian and NATO-member forces. Accidental weapons discharge, mistaken identification, or unauthorized actions by serving military personnel may qualify if they directly constitute an armed engagement involving the opposing side. The destruction of uncrewed drones, missiles, or other unmanned equipment does not qualify unless a human death directly results. An interception of a Russian missile or drone by NATO forces that causes civilian fatalities through falling debris qualifies only if the evidence establishes a direct armed engagement between the two sides and attributes the death to that engagement; merely intercepting a weapon aimed at Ukraine does not automatically qualify. Russian private military companies, intelligence operatives, and other irregular formations count only when credible evidence establishes that the personnel involved were formally incorporated into, or directly operating as part of, Russia's armed forces during the incident. Equivalent standards apply to NATO-member forces. Ukraine is not treated as a NATO member unless it formally joins the alliance before the qualifying incident. NATO membership is determined as of the incident date. A NATO member's military personnel fighting alongside Ukrainian forces can qualify, but only if they are operating in an official military capacity rather than as private volunteers or former soldiers. Neither a formal declaration of war nor invocation of NATO Article 5 is required. The geographical location of the incident does not affect eligibility.
More public forecasts
NATO x Russia military clash by December 31, 2026?
14%chance of yes
Prolepsia forecast
Will the Sudanese Armed Forces and the Rapid Support Forces agree a nationwide ceasefire before 31 December 2027?
20%chance of yes
Prolepsia forecast
Will NATO's 2027 estimates show every member spending at least 2% of GDP on defence?
1%chance of yes
Prolepsia forecast
Will EU imports of Russian LNG in 2027 be less than half of the 2026 level, according to Eurostat data?
96%chance of yes
Prolepsia forecast
Will the revised EU foreign investment screening regulation apply in all member states before 31 December 2027?
<1%chance of yes
Prolepsia forecast
Will a member of Vox hold a ministerial post in Spain's national government before 1 January 2028?
71%chance of yes
Prolepsia forecast