38 doomsday predictions since 1524. None happened. Four killed people.
Every dated end-of-world prediction we could source, what it cost the people who believed it, and the four-step pattern every single one of them follows.
We build a preparedness product. The strongest argument we have runs directly against the thing that sells best in our category, so we are going to make it anyway.
We went looking for every end-of-world prediction we could find that named a specific date and had a documented source. We found 38 of them, spanning 1524 to 2025. Not one happened. That is a success rate of zero across five centuries, and it includes predictions from astronomers, mathematicians, pastors, physicists, mystics and at least one hen.
The four-step pattern
Sociologist Landon Schnabel named the shape of it while commenting on the 2025 RaptureTok cycle: prediction, failure, reinterpretation, new prediction. Once you see it, you cannot stop seeing it.
Harold Camping predicted the Rapture for September 6, 1994. When it did not happen he moved it to September 29, then to October 2. Seventeen years later he predicted May 21, 2011, backed by a billboard and RV campaign reportedly costing around one hundred million dollars. When May 21 passed he announced it had been a "spiritual" judgment and moved the physical end to October 21.
William Miller ran March 21, 1843, then recalculated to October 22, 1844. The second failure was so total that it has its own name: the Great Disappointment.
David Meade ran September 23, 2017, then April 23, 2018. Ronald Weinland ran 2011, then 2012, then 2019. Leland Jensen ran 1980, then 1986, then 1987. Joshua Mhlakela ran September 2025, and when that passed, moved to October dates on the Julian calendar.
The revision is not a flaw in the belief. It is the mechanism by which the belief survives contact with reality.
What it cost
This is the part that gets left out of the listicles.
In March 1997, thirty-nine members of Heaven's Gate died in Rancho Santa Fe, California, believing a spacecraft trailing Comet Hale-Bopp would collect them. Marshall Applewhite had told them suicide was "the only way to evacuate this Earth."
In March 2000, roughly 778 members of the Movement for the Restoration of the Ten Commandments of God died in Uganda in a church fire and a series of linked poisonings, after the predicted end failed to arrive.
In 1975, Jehovah's Witnesses were counselled against pursuing higher education because Armageddon was expected that September. In 2011, followers of Harold Camping spent their savings. In 2025, people in the RaptureTok cycle sold homes, sold cars, and quit jobs.
The 1524 London flood prediction sent roughly 20,000 people fleeing their homes, and one merchant built an ark on high ground above the Thames. Nobody died. Everybody lost a week and looked foolish. That is the good outcome.
The ones that are not what you have been told
Three of the most-circulated predictions are misreadings, and correcting them matters more than listing them.
Isaac Newton did not predict the world ends in 2060. In private manuscripts around 1704 he computed 2060 as the date the 1,260-day prophetic period of Daniel concludes. His own stated purpose was the opposite of a prediction. In his words: "This I mention not to assert when the time of the end shall be, but to put a stop to the rash conjectures of fancifull men." He was establishing a floor, a date before which the world definitely would not end, specifically to shut down date-setters. He never published it. Every headline that reads "Newton predicted 2060" inverts him. The Newton Project's own statement is here.
Baba Vanga has no documented predictions. Wikipedia's article on her states flatly that "there are no written records about her alleged predictions," and that people close to her have said she never made many of those attributed to her. Every dated Vanga prediction in circulation, the 2025 European war, the 2028 Venus expedition, the 2043 caliphate, the 3005 Mars war, is a later attribution with no traceable source. The commonly cited Vanga date for the end of the world, 3797, is lifted directly out of Nostradamus's 1555 preface. Somebody copied it and changed the name.
Yellowstone is not overdue. This one is contradicted by the agency that monitors it. The USGS: "Yellowstone is not overdue for an eruption. Volcanoes do not work in predictable ways and their eruptions do not follow predictable schedules." The three caldera eruptions average roughly 725,000 years apart, which would put an "average" one about 100,000 years in the future, and USGS calls even that arithmetic meaningless. The magma chamber is 5 to 15 percent molten. USGS says so directly.
The one coming up in seven weeks
The nearest dated prediction on our calendar is November 13, 2026, and it is the most interesting entry on the entire list because it is real science that was never meant seriously.
In November 1960, Heinz von Foerster, P. M. Mora and L. W. Amiot published a paper in Science titled "Doomsday: Friday, 13 November, A.D. 2026." They fitted a hyperbolic curve to historical world population and found it goes to infinity on that date. It was a peer-reviewed joke, written to demonstrate that naive extrapolation produces absurd results. Von Foerster reportedly chose the date partly because it was his 115th birthday.
The model was falsified within a decade. Global population growth rate peaked in the 1960s and has declined ever since. But the paper is real, it is in Science, and the headline is irresistible, so every few months a fresh wave of coverage announces that a physicist calculated the exact date humanity ends, with the satire carefully removed.
We have it on the calendar with a live countdown and a label that says No basis. Both things are true at once, and that is the whole editorial position of this company.
So why build a preparedness product at all
Because the case for planning never needed a prophecy, and the prophecies are actively in the way.
Hurricanes do not consult a calendar. Neither do grid failures, wildfires, chemical spills, boil-water notices, or the ninety minutes when a highway closes and you discover your family had four different assumptions about where to meet. Those things have base rates you can actually look up, and none of them care what year it is.
The honest pitch is smaller and less exciting than the one our competitors make. You are not preparing for the end of the world. You are preparing for a Tuesday that goes badly, and the difference between a family that has thought about it and one that has not is mostly a matter of who spends the first two hours arguing.
Anyone selling you a date is selling you a date. We put all 38 failures on one page so you can check the base rate yourself.
Sources
Every prediction in this post is documented on our prediction calendar with a direct link to a primary or reliable secondary source. The dataset is also published as machine-readable JSON at /data/predictions.json under a permissive reuse policy. If you cite this research, a link back is appreciated but not required.