← Longevity & Supplement Guides
Humans could live to 194. The number that matters is 100.
The number surfaced in mid-July and spent the first week of August getting rounder. One hundred and ninety-four. Sometimes 156. Occasionally 470, in the write-ups that had stopped reading by the second paragraph. A group at Skoltech and AIRI in Moscow published a model in npj Aging asking how long a human body could last if you switched off every mechanism of ageing except one — the slow accumulation of mutations in your own cells — and the internet did what it always does when a large number stands next to the word lifespan.
The paper is careful and genuinely interesting. It is also being read almost exactly backwards. The headline reads 194 as a destination. The model arrives at it by subtraction.
What the model actually did
The method is called incremental modelling, and it is the opposite of a prediction. You start with a population survival model for a hypothetical organism that does not age at all. Such a creature still dies — of accidents, infections, bad luck — and in this framework its median run comes out at 1,759 years. Then you switch ageing back on, one mechanism at a time, and watch the number fall.
Switch on somatic mutations in the cells that the body never replaces, and 1,759 collapses to 156. Integrate the arithmetic across all the critical organs and you land at a median of 146 to 194 years. That is the figure that went around the world.
Notice what kind of number it is. It is not a forecast of how long anyone will live. It is the ceiling that would remain after every other mechanism of ageing had already been solved — proteostasis, mitochondrial function, epigenetic drift, the lot. Not one of them has been. The 194 is what is left when you delete the reasons people actually die and then measure the wall behind them.
The bottleneck is the finding, not the ceiling
The part of the paper worth keeping is the asymmetry between tissues, and almost nobody printed it.
Proliferating tissue turns out to be remarkably robust. In the model, the liver keeps replacing damaged cells so effectively that mutation-driven decline is more or less neutralised; it would go on working for thousands of years. The constraint sits somewhere else entirely — in the post-mitotic cells, the ones you finish building in early life and then simply keep. Neurons. Cardiomyocytes. "Neurons and cardiomyocytes turned out to be the main limiting factor," as the lead author put it.
There is something clarifying in that. The tissue that restocks itself is fine. The tissue you were issued once is the problem. Every mutation in a heart-muscle cell is permanent inventory, and the organ has no returns policy.
The authors are not making the claim the headlines made
Read the researchers' own commentary and the argument gets narrower and better. Somatic mutations, one of the co-authors noted, contribute substantially to ageing but "cannot by themselves explain the observed mortality." The gap between 146–194 and the eighty-odd years people actually get is filled by all the mechanisms the model switched off on purpose. Which means the useful conclusion is not we could live to 194. It is: mutations account for this much of the problem, and the rest of the hallmarks account for comparably more, so here is where to spend the next decade of research.
That is a paper about research priorities wearing the costume of a paper about immortality. Any model that removes the causes of death you are statistically going to die of will return a large number. The size of the number is a property of the deletion, not of the body.
The ceiling has been argued about for thirty years and never resolves
None of this is new terrain. In 2016, Dong, Milholland and Vijg published an analysis in Nature arguing that the maximum reported age at death had stopped rising around the mid-1990s and that human lifespan runs into a natural limit somewhere near 115 years. Two years later, in Science, Barbi and colleagues examined 3,836 Italians who had reached the age of 105 and found that their mortality hazard essentially stopped climbing — a plateau. If your annual risk of dying no longer rises with each birthday, there is no arithmetic ceiling at all, only increasingly improbable luck. Subsequent reanalyses using different longevity databases found rising hazards rather than flat ones, and the dispute continues.
Meanwhile the observed record — 122 years, set in 1997 — has stood unbeaten for twenty-nine years.
What is actually being fought over is the shape of the extreme tail of a distribution, decided by a few hundred very old people whose birth records are notoriously hard to verify. It is a good argument. It is also, for every single person reading this, entirely decorative.
The number that describes real people is 100, and most will not see it
Here is the study that should have been the news. In 2024, Olshansky and colleagues published an analysis in Nature Aging of the ten longest-lived populations on earth — Japan, South Korea, Australia, France, Italy, Switzerland, Sweden, Spain, Hong Kong and the United States — across the years 1990 to 2019.
For most of the twentieth century, each decade added roughly three years to average life expectancy in wealthy countries. Since 1990, those same populations have gained about six and a half years in total, and the rate of improvement decelerated in almost all of them. The authors then ran the projection forward: of children born in 2019 in Hong Kong, the longest-lived population measured, an estimated 12.8% of girls and 4.4% of boys would reach the age of 100. In the United States: 3.1% and 1.3%.
Sit those two stories side by side. One model says the wall is at 194. The other says that in the best-performing society we have ever built, roughly one girl in eight and one boy in twenty-three makes it to a hundred — and the curve that got us here has been flattening for a generation. The frontier was never the ceiling. It is the enormous, mostly empty space underneath it.
THE GAP
A modelled ceiling, and the distribution of actual people underneath it
Median lifespan — logarithmic scale
Projected to reach age 100, births of 2019 — axis 0–20%
Why the ceiling argument never settles
2016, Nature (Dong, Milholland, Vijg). Maximum reported age at death stopped rising around the mid-1990s; the authors put the natural limit near 115 years.
2018, Science (Barbi and colleagues). Among 3,836 Italians aged 105 and over, the mortality hazard flattened rather than climbing — which would mean no fixed ceiling, only steepening improbability.
Since. Reanalyses using other longevity databases report rising hazards after 105 rather than a plateau. The disagreement turns on a few hundred people whose ages are hard to verify — which is why three decades of work has not closed it.
Top panel: model outputs from Efimov and colleagues, npj Aging 2026, shown against two observed figures; bar lengths use a declared logarithmic scale because the values span 85 to 1,759 years, and every value is printed in its label. Bottom panel: linear bars on a 0–20% axis, from Olshansky and colleagues, Nature Aging 2024.
Three different numbers, permanently confused
Most of the noise around this story comes from collapsing three quantities that behave nothing alike.
Maximum lifespan is a record held by one individual. It moves when a single exceptional person dies, which is to say almost never. Life expectancy is the mean of a whole distribution, and historically it has been moved by clean water, vaccination, tobacco control and cardiovascular medicine — not by anything discussed on a wellness podcast. Healthspan is how many of those years you would describe as well, and it has been quietly falling behind: the global gap between living and living well now runs to more than a decade, and it is widest in the richest countries.
The affluent world has developed an unusual appetite for discussing the first number while living inside the third. A ceiling of 194 years is a fascinating fact about biology and a completely useless fact about Tuesday.
The bottleneck points somewhere unexpectedly practical
Take the model's actual finding seriously for a moment, though — because it is the one part of the story with a use.
What sets the limit, in this framework, is not the body in general. It is the two organs that do not restock: the brain and the heart. And those happen to be the two organs where the human evidence, as opposed to the modelling, is thickest and dullest. Cardiorespiratory fitness carries one of the strongest associations we have with a brain that ages more slowly. The people who hold on to their cognition into their eighties and nineties turned out, when the genomes were finally read, not to be carrying protective variants — whatever separated them was not written in advance.
None of that will take anyone to 194. It is simply that the model's bottleneck and the observational evidence point at the same tissue, and only one of them tells you what to do on a Wednesday.
The one thing worth doing this week: ignore the ceiling entirely and pick a fixed effort you can repeat — the same hill, the same stairs, the same twenty-minute route — and note how hard it feels and what your heart rate does at the top. Then repeat it weekly. What you are watching for is the relationship between pace and heart rate drifting in your favour over two or three months, which is the cheapest home proxy anyone has for cardiorespiratory fitness. Read the trend, not the day: a single reading is mostly sleep, heat and coffee. And if something feels wrong rather than merely hard, that is a conversation for your doctor, not a data point.
The bottom line
The npj Aging model is good work and its headline is a mirage. It says that if you solved every other mechanism of ageing, permanent mutations in your non-renewing cells would still stop you somewhere between 146 and 194 years — and its own authors point out that those other mechanisms are doing comparably more of the damage right now. Meanwhile the population data says improvement has been slowing since 1990 and that in the healthiest society on record, most people still will not see a hundred.
So the ceiling is not the constraint. The distance to it is. Whatever raises the floor — sleep, movement, food, the people you see — remains unglamorous, unfinished and entirely available, which is a less exciting sentence than "humans could live to 194" and a considerably more useful one. Use numbers to correct fantasy, not to replace experience.
