Skip to main content
We’re Edging Closer To Nuclear War

Welcome to Bayesian Update, an attempt to take a step back and figure out what’s changed in the world since before President Trump was elected. In each piece, I’ll pose a high-stakes question with a numerical answer and ask a panel of experts to give their estimates and describe recent trends.

The nuclear football — a black briefcase containing an illustrated menu of doomsday scenarios — follows President Trump everywhere he goes. Like every U.S. commander-in-chief since John F. Kennedy, Trump has the sole authority to empty the American nuclear arsenal on any target, at any time, for any reason. James Mattis, his secretary of defense, must authenticate the order before it reaches the Pentagon, but should Mattis refuse to do so in an attempt to prevent missiles from launching, Trump can simply fire him on the spot and replace him with someone who will carry the order out. “There is no procedural or institutional mechanism that can stop a president from giving an order to use nuclear weapons,” said Stephen Schwartz, editor and co-author of “Atomic Audit: The Costs and Consequences of U.S. Nuclear Weapons Since 1940.”

You can exhale, though: Most nuclear security experts I spoke to are not particularly worried by this aspect of the Trump presidency. They said that the risk of civilian-targeted nuclear weapon use has ticked up since 2015, but the causal pathway is a bit subtler than itchy fingers on the metaphorical red button. “I don’t know how this plays out,” said Rachel Bronson, executive director and publisher of the Bulletin of Atomic Scientists. “But he’s moving us into a much more uncertain time.”

The trouble is, nuclear risks are hard to measure quantitatively. The small sample size (two bombs dropped, ever) and rapidly changing technological and diplomatic contexts don’t exactly lend themselves to simple mathematical modeling. While such models do exist, they are “mainly an exercise in structuring one’s thinking, not something that would provide a ‘right’ answer,” according to Matthew Bunn, a professor at the Harvard Kennedy School.

But just because we can’t model our way to an exact answer doesn’t mean we should throw up our hands and move on. Since so many lives are at stake, even a tiny increase in the probability that nuclear weapons will be used is a really big deal, and that remains true even if our best predictions are somewhat imprecise.

Academics and diplomats who spend their careers studying nuclear weapons have a pretty good conception of the nature and magnitude of the risks — their back-of-the-envelope estimates are as good an answer as we have. And while some experts disagreed on the details, everyone I spoke to painted the same general picture.

In short, a nuclear strike on a civilian target could realistically happen in one of two ways: Either

  1. tensions between two nuclear states rise to the point where a single miscommunication or technical failure could trigger a launch; or,
  2. a terrorist organization could acquire nuclear weapons capabilities.

So how likely is either scenario?

The nuclear football as it appeared during the administrations of Bill Clinton, George W. Bush and Ronald Reagan.

Getty images


State use of nuclear weapons is more likely than you think

On the state side, there are a number of ongoing conflicts that could, in theory, go nuclear at any time. “Increasingly, some regional powers are relying on nuclear weapons for their day-to-day security against conventional conflict,” said Vipin Narang, author of “Nuclear Strategies in the Modern Era.” “If they think that a conventional invasion is coming — whether it is or not — they may be worried that the nuclear forces that they rely on for their survival might be threatened … there may be what’s sometimes called a ‘use it or lose it’ situation.”

The conflict that topped experts’ list of clashes to be concerned about is India-Pakistan. Both states have developed nuclear weapons outside the jurisdiction of the Non-Proliferation Treaty, both states have limited capabilities, which may incentivize early use, and both states — though their public doctrines are intentionally ambiguous — are known to have contingency plans involving nuclear first strikes against military targets.

Then there’s North Korea, whose recent missile tests have brought renewed attention to the state’s nuclear weapons program, which has spurred international trade sanctions. The Korean War never officially ended, so North Korea is still technically facing the threat of a U.S.-backed South Korea, and nuclear weapons remain central to North Korea’s national defense strategy. Some experts believe that the seemingly erratic behavior of the Kim regime is in fact strategic: If you’re handcuffed to your adversary on top of a cliff, dancing erratically near the edge is a smart way to extract concessions.

Beyond these two clear danger zones, several experts cited U.S.-Russia or Iran-Israel as distant third-place threats to go nuclear, with one suggesting that U.S.-China could heat up in coming years as the situation in the South China Sea develops.

In any of these active conflicts, we shouldn’t necessarily expect that fear of mutually assured destruction will save the day. We can’t say with any confidence how likely a nuclear conflict is because we don’t know what a total war between two nuclear states would look like — we’ve never had one. “You’d like to hope if there was some sort of conflict started, it would remain limited and conventional until people could tamp it down,” said David Wright, co-director of the global security program at the Union of Concerned Scientists. “But you can certainly imagine ways it would start to get out of control.”


Nuclear terrorism is plausible, but difficult to pull off

Similarly, just because there’s never been a nuclear terrorist attack doesn’t mean that it will never happen. In theory, if a non-state actor got ahold of enough fissile material — the active ingredient in nuclear weapons — it would be relatively easy for them to assemble and detonate a bomb, according to Robert Rosner, former chief scientist and laboratory director at Argonne National Laboratory. “You’d need some physicists who know what they’re doing,” Rosner said. “But based on what’s available in the public literature, you could go ahead and make a uranium bomb.”1 Detection and prevention at this point would be very difficult, Rosner says — a weapon could be assembled in a garage and smuggled in a standard box truck.

Fortunately, fissile material is hard to come by. The processes used by states to develop fissile material — a diffusion plant or farm of specialized centrifuges for enriched uranium, a specialized reactor for plutonium-239 — would be prohibitively expensive for a non-state actor. Plus, due to their size (dozens of acres), these facilities are highly conspicuous and would likely be identified and destroyed before a terrorist cell could refine enough material to pose a threat.

A terrorist with nuclear ambitions, then, would have to acquire existing fissile material from one of the nine nuclear states, which could happen in one of two ways. First, there’s open theft, either of fissile material or of a fully assembled weapon. This would likely require a firefight, according to Rosner — nuclear facilities have armed guards2 — which would alert authorities to the presence of a threat. Second, which is the likelier possibility according to several of the experts I talked to, is through the assistance of an insider: A double agent with terrorist sympathies could infiltrate a state’s nuclear apparatus and simply deliver a weapon to a non-state actor.

On both counts, Pakistan again emerged as the consensus pick for the No. 1 cause for concern, largely due to its instability. “If the Pakistani state does collapse, it probably wouldn’t collapse in one big bang, but slowly become more and more dysfunctional,” said Ramamurti Rajaraman, professor emeritus of physics at Jawaharlal Nehru University. “If the dysfunctionality also happens in the nuclear weapons security apparatus of Pakistan … that I see as the biggest danger.”

Finally, an act of nuclear terrorism would require the existence of a non-state actor that had both the organizational sophistication and the military ambition to entertain the prospect of nuclear violence. “I would say at the moment Al Qaeda and its various branches and ISIS are the main terrorist groups where … it’s at least within the realm of the plausible that they’d be able to do this,” said Bunn. “Compared to 2015, I’m at least modestly less worried about the Islamic State, in that they seem to have turned to very unsophisticated attacks … and are under huge pressure militarily.”

Though most experts I spoke to considered both state and non-state risks to be serious and worthy of attention, a clear majority (four of the five who were willing to choose) thought that state use of nuclear weapons was more likely than use by terrorists. “I’m more worried about a nuclear state,” said Wright. “They have large numbers of these things; they’re worked into the war plans. They practice using them.”


If a state uses a nuclear weapon, it’ll probably be by accident

When you imagine state use, though, don’t think of a red-faced Trump or Kim launching a petty revenge strike. “Nobody’s going to wake up one morning and say, ‘Gee, today would be a really great day for a nuclear war,’” Bunn said. These scenarios account for a tiny sliver of the probability that nuclear weapons will be launched at civilian targets.

The real risk, embarrassingly enough, is accidental strikes. Amidst the chaos of an international crisis, global catastrophe could arise from a mere technological error — it only takes one falling domino to trigger an avalanche of self-defense responses, Bronson said. “We know the history. We know that conflict has the potential to escalate quickly,” she said. “When we have huge arsenals on high alert, accidents can happen that can be very dangerous.”

If this sounds more like “Dr. Strangelove” than reality, you may want to take a spin on the Wheel of Near Misfortune, where the Union of Concerned Scientists shares stories of instances where the world only narrowly avoided a nuclear strike. There have been a shocking number of close calls, where a faulty reading or hardware malfunction nearly provoked a nuclear response. Now swallow this: There’s nothing built into the system that has caused the coin to always come up heads so far. “We were prepared — and are still prepared — to use [nuclear] weapons at a moment’s notice,” said Schwartz. “The fact that we didn’t is not necessarily proof that the system works so much as proof that we got very lucky.”

If anything, we have reason to believe we won’t always be so lucky. “All of those incidents occurred during peacetime, so there were lots of indications that this is not normal,” said Schwartz. “If those kinds of incidents happen during a crisis, where everything is ratcheted up a few notches, and you’re already feeling kind of edgy, then not only are you perhaps convinced that it’s a real attack — as opposed to a glitch of some kind — but your system is geared to respond all the more rapidly.”

Wright gave a more specific example: “If you couple … a conventional conflict that is escalating with an attack on U.S. satellites, so that the U.S. loses important communications and surveillance systems, those war games frequently go nuclear.”

Three recent international trends have raised the risk

Humanity’s best recourse, if we (prudently) assume that accidents are inevitable, is to back away from the edge of the cliff until we can afford a stumble or two without falling off. But we have not done this — quite the opposite. The experts I spoke to pinpointed three interlocking trends that they believe have brought us closer to the brink than we were in 2015.

“Nuclear disarmament is the only way to get rid of the threat,” said Kane. “That is simply not happening right now.”

First, the last two years have seen a sharp resurgence in ethnic and religious nationalism across the West, with several countries deprioritizing postwar liberal values of international cooperation, pluralism, and freedom of trade and migration in pursuit of national might and a coherent national identity. Marine Le Pen, president of the right-wing French nationalist party National Front, described these competing visions last November: “The model that is defended by Vladimir Putin is radically different than that of Mr. Obama. As for me, the model that is defended by Vladimir Putin — which is one of reasoned protectionism, looking after the interests of his own country, defending his identity — is the one that I like.”

Opinions vary on the domestic merits of this political shift, but the experts I spoke to were unanimous in condemning this strain of anti-globalism and anti-multilateralism from the perspective of nuclear security. The main concern is that nationalist governments might “take measures to increase their survivability in ways that would not be particularly conducive to global security,” said Narang.

“There is a reassessment of the security politics,” said Angela Kane, former high representative for disarmament affairs at the United Nations. “When you look at [the election of President Trump], particularly coupled with the Brexit decision last year,” and the growing electoral strength of far-right parties “in the Netherlands, France, Germany — all of this brings an instability into the situation that people are greatly worried about.” International peace, after all, rests on a number of treaties and assumptions that are now being called into question.

The recent political shifts are “certainly not conducive to the architecture we’ve worked within for the last 70 years or so,” Bronson said. “I do think this is a much more dangerous world.”

Second, the world’s strongest military power, under its new, more nationalist government, has signaled interest in renegotiating the security agreements that help ward off war — nuclear and conventional — in Europe and East Asia. “All of a sudden there is a questioning of the commitments that the United States has made and the leading role that the United States has played in multilateral diplomacy,” said Kane. “It hasn’t been said so publicly, but … there’s been a realization that maybe the Europeans need to do a bit more for their own defense.”

“We’ve seen this movie before,” said Narang. “The Eisenhower administration went to tremendous lengths to establish essentially nuclear sharing agreements with [West] Germany … to stop them from getting the weapons … so that we and we alone could control nuclear use and escalation.” Removing the nuclear umbrella and encouraging allies to go it alone can only complicate the picture. “The more countries that have nuclear weapons, the more nuclear weapons there are in the system, the more actors have the ability to use them … the probability of use just accumulates,” he said.

Bunn put it bluntly: “It would be disastrous for the U.S. to withdraw its protection from these countries.”

If there’s any cause for optimism on this front, it’s Secretary of Defense James Mattis, who in 2016 criticized both President Obama and then-candidate Trump for their shared view of American allies as military “free riders.” On his first international trip as secretary of defense, Mattis went to Japan and South Korea to reassure leaders that American nuclear commitments remained strong. “There is apparently already a repositioning of the United States which is not exactly aligned with the statements that President Trump made initially,” said Kane. “That, to my mind, is also significant.”

Whether Mattis can check the president’s instincts and preserve the “Washington playbook,” though, remains to be seen. “I do find comfort in the fact that Mattis is extremely experienced and has a lot of respect,” said Bronson. “But Mattis is one voice in an administration with a lot of competing perspectives. It’s unclear how it will eventually be organized, or what the administration’s worldview will be.”

The third trend is, in the context of nuclear weapon use, perhaps the most significant: “The disarmament process has come to a halt,” said Rajaraman. The assertion that the U.S. will not renew the New START treaty, a bilateral agreement that limits Russian and American stockpiles; the pending review of the 2015 deal that curbs Iran’s production of fissile material; Trump’s signals to other nuclear powers that the U.S. intends to expand and modernize its arsenal3 — this is not just talk. These are concrete actions that work directly against the program of nuclear disarmament, which has been progressing in fits and starts since the end of the Cold War.

This matters. International conflicts will flare up and fade away, but weapons stockpiles remain the underlying source of all nuclear dangers, state and non-state. “Nuclear disarmament is the only way to get rid of the threat,” said Kane. “That is simply not happening right now.”

Experts agree that we could reduce stockpiles significantly — and thus reduce the risk of their use significantly — without reducing nuclear weapons’ power as deterrents.4 Still, nuclear states are hesitant to move in this direction, for fear that moving toward disarmament would be interpreted as a sign of weakness. Multiple experts I spoke to expressed frustration that these political considerations should outweigh the strong consensus of people who study nuclear security and deterrence dynamics. “A lot of arguments are made that actually don’t make any sense,” said Wright. “We’ve done a lot of work on this.”

It’s unclear yet whether the recent moves toward rearmament represent a temporary blip or a turning point. This is the 1,000-kiloton question. If weapons stockpiles continue to grow, the per-year risk of civilian-targeted nuclear weapon use will only increase. “Hopefully nobody is crazy enough to drop one,” said Rajaraman. “But nobody has the guts to get rid of them. I think it’s going to go on like this until something stupid happens.”


  1. A plutonium bomb, according to Rosner, is likely beyond the capability of even a terrorist group with physics expertise: “There are too many things you would need … which are not widely available.”

  2. Rosner stressed that, while the theft would still be difficult, a terrorist could in theory acquire radioactive material (not to be confused with fissile material) from “soft targets” like hospitals. This could be used to create a “dirty bomb,” which is a traditional bomb rigged with radioactive material that could render the blast area uninhabitable for decades.

  3. The United States was already moving away from disarmament, with Obama’s nuclear modernization program, before Trump took office. Trump has also expressed interest in expanding and modernizing the arsenal, though it’s unclear whether his approach will be a continuation of Obama’s program or something else.

  4. Larger stockpiles do enhance deterrence, but only to a point. Bronson said that the United States’ strategic arsenal (estimated to be 2,080 weapons) could be reduced “to 1,000 without affecting deterrence.” Wright added that all land-based missiles could be taken off high alert: Nuclear submarines, which are difficult to locate and target, are sufficient demonstration to adversaries that the U.S. could “absorb and ride out a first strike and … have enough that you’d be able to respond.”

Milo Beckman is a freelance writer for FiveThirtyEight. His work can be found at He also constructs crossword puzzles for The New York Times.