Ten Near-Misses That Almost Started a Nuclear War

On October 5, 1960, the American early warning radar at Thule, in northern Greenland, sent an alarming report to North American Air Defense Command headquarters in Colorado Springs. According to the system, dozens of Soviet missiles were on their way. NORAD went to its highest level of alert.

The radar had been fooled by the moon. As it rose over Norway, the computers interpreting the radar data misread it as a massive missile launch. What saved the day was context, not technology. Soviet leader Nikita Khrushchev was in New York City at the time, attending the United Nations. Officials reasoned that Moscow was unlikely to launch an all-out attack while its own leader sat in Manhattan, and that doubt bought time until the error became clear.

The episode set a pattern that runs through this file. Warning systems built to buy decision time can also manufacture false emergencies, and the people reading them have to judge, quickly and under pressure, whether to believe the screen. In 1960 the deciding clue was not a sensor reading at all. It was a newspaper fact about where one man happened to be standing that week.

On November 24, 1961, the staff at Strategic Air Command headquarters in Omaha suddenly lost contact with the early warning radar at Thule, Greenland. At the same moment, their line to NORAD in Colorado went dead too. Those links were supposed to be independent of each other. Losing both at once made little sense as an accident, and it looked a lot like the first moments of a coordinated attack meant to blind American defenses. SAC ordered its entire alert force of bombers to prepare for takeoff.

The alert was called off after a U.S. bomber flying over Thule made radio contact with the radar site and reported that all was normal. The cause turned out to be mundane: a single AT&T switching station in Colorado had failed. The phone company had assured the government that these critical circuits had backup routes. In practice, they all ran through the same point.

Nobody was hurt, and no bomber left the ground. But the episode showed how the machinery of nuclear deterrence could read an ordinary equipment failure as an act of war, and how much depended on someone finding a way to check before acting.

This one was an accident, not a false alarm, and it happened on American soil. On January 24, 1961, a B-52G from Seymour Johnson Air Force Base developed a fuel leak in its right wing while on airborne alert. Ordered back to base, the bomber lost control and broke apart near Faro, about 12 miles north of Goldsboro. Three of the eight crew members died.

Two Mark 39 thermonuclear bombs, each with a maximum predicted yield of about 3.8 megatons, fell free. On one, the parachute opened and it landed nearly intact, but the crash had started parts of its arming sequence. What stopped it was its arm/safe switch, which remained on safe. The second bomb slammed into a muddy field. Its switch was found turned to arm, but impact damage meant it never received the electrical charge needed to fire. Crews dug to 42 feet and never recovered part of it. The landowner was paid $1,000 for a permanent easement over the spot.

In 1969, a Sandia supervisor named Parker Jones wrote that a single simple low-voltage switch had stood between the United States and catastrophe. Engineers still disagree about how close it came; some argue the safety system worked as designed. North Carolina put up a historical marker in 2012.

On the same day that B-59 was cornered, October 27, 1962, an American U-2 spy plane flying out of Eielson Air Force Base in Alaska went badly off course. Its pilot, Captain Charles Maultsby, made a navigation error, reportedly made worse by the shimmering aurora borealis, and strayed roughly 300 miles into Soviet airspace over the Chukotka Peninsula.

Soviet MiG fighters scrambled to intercept him. To protect the U-2 as it headed back toward Alaska, the U.S. sent up F-102 fighters armed with nuclear-tipped air-to-air missiles. At the peak of the Cuban Missile Crisis, American and Soviet fighters were converging in the Arctic, with nuclear weapons on board one side.

Maultsby made it back without a shot fired. The incident might have stayed a footnote if Khrushchev had not raised it himself. In his letter to Kennedy the next day, he complained that an American reconnaissance plane had crossed into Soviet territory in Chukotka. He warned that, with both sides on high alert, such an intruder could easily have been mistaken for a nuclear bomber, which might have pushed the Soviet Union toward a fateful step. The letter, now published by the State Department's historians, shows the danger was on his mind at the time.

On the morning of November 9, 1979, computers at NORAD suddenly showed a large-scale Soviet missile attack on the United States. The response was immediate. Missile crews went to their highest state of readiness, bombers prepared to take off, about a dozen fighter interceptors were launched, and the National Emergency Airborne Command Post took off, though without President Carter aboard.

Within about six minutes, the alarm began to unravel, because satellite data did not confirm any launches. The cause was human error: a tape containing an exercise scenario had been loaded into an operational computer, so the system displayed a simulated attack as if it were real. A widely retold story has National Security Adviser Zbigniew Brzezinski woken in the middle of the night and moments from calling the president. The National Security Archive has cautioned that this account, from a later memoir, may blend this event with a different false alarm in June 1980.

Moscow noticed. Within days Soviet leader Leonid Brezhnev sent a message to Washington calling the incident fraught with tremendous danger, especially after press reports that top U.S. leaders had not been told at the time. Washington replied that the Soviet message was inaccurate and unacceptable, but the episode fed growing worry about the reliability of early warning computers.

Just seven months after the training tape scare, it happened again, twice. On June 3, 1980, in the middle of the night, U.S. warning displays began showing Soviet missile launches. On June 6 they did it again.

Each time, the alerts moved further than anyone would like. Bomber and tanker crews were ordered to their stations and started their engines. The National Emergency Airborne Command Post, the flying headquarters meant to keep command of U.S. forces during a nuclear war, taxied for takeoff. Then, as no further signs of an attack appeared, the alerts were called off. The Defense Department later traced the problem to a failed integrated circuit, a computer chip that cost 46 cents.

The June incidents came at a tense moment, about six months after the Soviet invasion of Afghanistan had chilled relations between the superpowers. Coming so soon after the November 1979 tape error, they made it hard to argue that false alarms were rare flukes. Marshall Shulman, a senior State Department adviser, wrote that false alerts of this kind were not a rare occurrence. They also left a lasting lesson about nuclear command systems: the most dangerous component can be a cheap part nobody thinks about, buried deep inside a machine that everyone trusts.

The Cold War was over, but its machinery was not. On January 25, 1995, a team of Norwegian and American scientists launched a four-stage Black Brant XII research rocket from the Andøya rocket range off northwestern Norway to study the northern lights. It climbed to about 903 miles (1,453 km) before falling near Svalbard, roughly 24 minutes later, just as planned.

Russian early warning radar saw something more sinister. The rocket's flight profile resembled that of a missile launched from an American submarine, and Russian nuclear forces went on high alert. The alert went up the chain, and the nuclear briefcase known as the Cheget was brought to President Boris Yeltsin. He activated it, the first known time a leader of a nuclear power has done so.

Russian observers soon determined that the rocket posed no threat, and no retaliation was ordered. The frustrating part is that the launch had been announced. Norway had notified about 30 countries in advance, including Russia, but the notice reached the Foreign Ministry and never got to the military officers watching the radar screens.

On September 26, 1983, less than four weeks after the Soviet Union shot down a Korean airliner, Lieutenant Colonel Stanislav Petrov was the duty officer at a command center for Oko, the Soviet satellite early warning system. The screens reported a launch of an American intercontinental missile. Then another, and more, until the system showed five.

Procedure was to pass the warning up the chain, where it could feed a decision to retaliate. Petrov had minutes. He judged it a false alarm, for several reasons: a real American first strike would surely involve far more than five missiles, the new satellite system was not fully trusted, and ground-based radar showed nothing. He reported a malfunction instead. He was right. Investigators later concluded that sunlight glinting off high-altitude clouds had fooled the satellites.

Petrov was not rewarded. He was reprimanded for gaps in the logbook and later retired early. The world learned of the incident only in 1998, from the memoir of a retired Soviet air defense general. Analysts still debate how decisive his choice was. Some argue Soviet leaders would have acted on such a warning; others, such as researcher Pavel Podvig, note that more layers of decision sat above him. Petrov received the Dresden Peace Prize in 2013 and died in May 2017.