The death of a big star, much more massive than our sun, usually proceeds like this: After millions and millions of years of shiny existence, the star starts to run out of hydrogen. Without this fuel, the star can’t power the nuclear fusion that produces its light. Its core shrinks and heats up, spawning heavier and heavier elements until mostly iron remains. Within a second, the core collapses and sends star material flying in a spectacular light show—a supernova—that fades after several months. The dead star leaves behind a neutron star, a very dense object, or a black hole, the light-gobbling lurkers of the universe.

What a star isn’t supposed to do, however, is stay alive.

A team of international astronomers announced Wednesday the discovery of a supernova that occurred in 2014 and continued to erupt for more than 600 days, making it the longest such explosion ever observed. The light of typical supernovae usually lasts only 100 days, and anything more than 130 days is extremely rare. When astronomers dug into archival data, they found evidence of another supernova in the same location in an image from 1954. The star responsible for both explosions, the researchers say, somehow managed to survive a blast and explode again 60 years later.

The latest supernova was detected in September 2014 by the California Institute of Technology–operated Palomar Transient Factory, which is designed to spot new objects in the sky. The explosion, named iPTF14hls, was classified as an ordinary Type II-P supernova. Astronomers at the Las Cumbres Observatory, a global network of robotic telescopes, were studying the Palomar data in their search for supernovae at their peak. They took note of it, but moved on after they saw the supernova appeared to be fading. Later, in early 2015, Iair Arcavi, an astronomer at Las Cumbres, asked an intern, Andrew Wong, to go through the whole data set again. Wong found something unusual: iPTF14hls, once fading, had flared up. He asked Arcavi if this was normal.

“I said, absolutely not,” Arcavi said. “That’s very strange.”

After all, supernovae are a one-time deal. “A supernova gets bright and then fades,” he said. “It’s not supposed to get bright again.”

Arcavi thought iPTF14hls was more likely to be a nearby twinkling star that was swept up in a survey of supernovae. He and his colleagues decided to run some forensics on the object. They used telescopes to study its spectra, which provide information about its chemical composition, the speed of the material, the age of the blast, and other important properties. The results showed that this wasn’t a healthy, living star, but a supernova. “I was really surprised,” Arcavi said. “The last thing I expected to see was a spectrum of the most typical supernova, the most boring kind of supernova you can get.”

Arcavi directed Las Cumbres telescopes to stare nonstop at the mystery explosion. Every few days, new data came back. “As time went on, it just got more and more mysterious,” he said. The supernova was challenging every known theory about the end of a star’s life that astronomers have.

Over two years, the brightness of iPTF14hls varied over time in such a way that it seemed like it was exploding over and over again. The ejected material appeared to travel at the same speed over time rather than slowing down. The temperature of the explosion remained unchanged, too. One by one, every theoretical model for star death was thrown out, unable to explain all of the explosion’s bizarre attributes. The discovery of the image from 1954 further confounded the researchers.

The astronomers eventually settled on a theoretical model that describes the potential existence of a never-before-observed event with the strange name “pulsational-pair instability supernova.” In this model, very massive stars, as much as 100 times the mass of the sun, become so hot at their cores that they convert energy into matter and antimatter. An explosion follows, ejecting the star’s outer layers but leaving its core intact. More explosions could continue like this for decades before the core finally collapses.

But there are limitations to this explanation. iPTF14hls produced far more energy than the theory predicts it should. Arcavi said every scientist to whom he’s shown the findings is stumped. “And still now, even with the paper being published, we still don’t have any theory or any model that fully explains the observations of it,” he said. It even took some convincing for Nature to publish a paper that didn’t have a concrete answer for such a significant conundrum, he said. But to truly explain the supernova, astronomers around the world need to see the data and come up with a completely new model. Perhaps someone will even discover earlier images of the supernova in their own archival data, Arcavi said.

Sarah Sadavoy, an astronomer at the Smithsonian Astrophysical Observatory at Harvard University who was not involved in the study, called the long-lasting supernova a “strange event.” The biggest mystery in the findings, she said, is the detection of hydrogen at the scene of the cosmic explosion. “Hydrogen is found in the outermost layer of these massive stars, and as such, should be lost during the first burst,” Sarah said. “This particular event is just as powerful as most other supernova of this type, so it doesn’t make sense for it to still have hydrogen.”

But cosmic mysteries, while frustrating, are a good thing. “There are many things that we still don’t understand about the explosion mechanisms of massive stars and the associated element production,” said Anna Frebel, an astrophysicist at the Massachusetts Institute of Technology who studies stars. “Finding this object appears to cause more puzzling ‘problems’ since it’s never been seen before, but it’s those kind of challenges that help astronomers understand more about the death of stars.”

Arcavi said the discovery will force astronomers to reexamine the supernovae they’ve encountered in their observations. “Anytime someone reports a supernova of this type, we have to look at it again and check if it’s not one of these weird ones,” he said. “We don’t know how many of these we might have missed.”

Some researchers are calling the star responsible for iPTF14hls a zombie star, an undead being glowing in the cosmos. Arcavi said he has some mixed feelings about the nickname. The Walking Dead and many other depictions of the undead have shown that zombies, after considerable effort, can be killed. And when they’re dead, they usually stay dead, save for, perhaps, one final gasp when the unlucky human who took it out has turned away. This star, on the hand, won’t stop dying.

The supernova has begun to fade slowly in the last year, however. Astronomers hope that as the ejected material expands, the area may become transparent enough for powerful telescopes to glimpse the heart of the explosion. Hubble, the premier star-observing space telescope, will take a look next month. The future of iPTF14hls is unclear. Arcavi and his colleagues don’t know whether the star survived the explosion.

“It is weird that the current explosion is going on for so long, but that doesn’t necessarily mean that the star is still alive,” Arcavi said. “I wouldn’t bet my car on it either way.”