Nobody can give you a firm expiration date for the human species, but the range of informed estimates spans from disturbingly soon to comfortably far off. On the near end, some risk researchers judge the odds of an extinction-level catastrophe within this century at roughly one in six, driven mostly by advanced technologies we are still learning to control. On the far end, astrophysics gives Earth’s biosphere something like a billion more years before the brightening Sun makes the planet uninhabitable for complex life. Between those poles sits every threat you have heard of and a few you probably haven’t, each with its own probability and its own timeline. The honest answer is less about a single number and more about which risks dominate at which timescales.
What the Fossil Record Says About Species Lifespans
One way to approach the question is to ask how long mammalian species typically last before they go extinct or evolve into something different. The fossil record suggests that invertebrate species persist for roughly five to ten million years, while mammals average considerably less.1PubMed. Getting the measure of extinction A detailed analysis of North American fossil mammals puts the figure for larger mammals at about 2.3 to 4.3 million years, depending on the group, with an overall average near 3.2 million years.2PubMed. Species longevity in North American fossil mammals Homo sapiens has existed for roughly 300,000 years, so by the fossil-record yardstick we are still early in our expected run.
But this comparison has serious limits. Most mammalian species that went extinct did so because of environmental shifts, competition, or predation. Humans have largely escaped those pressures through technology and culture. We are also the first species capable of deliberately engineering our own destruction. The fossil baseline is useful for calibration, but it cannot account for the novel risks a technological civilization creates for itself.
The Next Hundred Years and Why They Worry Researchers Most
When experts in global catastrophic risk are surveyed, their estimates for the probability of human extinction before the year 2100 cluster uncomfortably high. Philosopher Toby Ord, in his book The Precipice, judged the overall likelihood of an existential catastrophe this century at about one in six. A separate 2008 survey of global catastrophic risk experts arrived at a comparable figure of roughly 19 percent.3PubMed Central. Extinction of the human species: What could cause it and how likely is it to occur? Both estimates are dominated by risks from emerging technologies rather than natural hazards. These numbers carry huge uncertainty and incorporate a fair amount of subjective judgment, and some critics argue the climate and pandemic components are inflated. Still, the broad agreement on order of magnitude across different expert groups is hard to dismiss.
This century matters disproportionately because we are in a period of rapid technological development where the tools to cause catastrophic harm are advancing faster than the institutions meant to govern them. Whether you find a one-in-six chance alarming or manageable depends on your risk tolerance, but for context, it is roughly the odds of losing at Russian roulette.
Nuclear War and the Famine That Follows
A full-scale nuclear exchange between major powers would kill hundreds of millions of people directly, but the extinction-level concern is what comes after. Even a regional nuclear conflict involving a small fraction of the world’s arsenal, less than 0.03 percent of current stockpiles, could loft enough soot into the upper atmosphere to block sunlight and trigger worldwide crop failures.4Eos. Regional Nuclear War Could Cause a Global Famine A full-scale war between the largest nuclear powers would produce far more soot and could plunge global temperatures by ten or more degrees for years, a scenario often called nuclear winter.
Would nuclear winter actually drive humanity to extinction, or would scattered survivors rebuild? Most researchers think total extinction from nuclear war alone is unlikely, because some regions and some people would probably escape the worst effects. But the combination of cold, darkness, crop failure, supply-chain collapse, and social breakdown could reduce the global population to a fraction of its current size, and recovery would be neither quick nor guaranteed. The risk is less “everyone dies at once” and more “civilization collapses so thoroughly that rebuilding becomes questionable.”
Engineered Pandemics and the Biotechnology Threat
Natural pandemics have killed enormous numbers of people throughout history, but they have never come close to wiping out the species. The reason is simple: pathogens that kill their hosts too quickly tend to burn out before they can spread everywhere, and genetic diversity in a large population means some individuals will be naturally resistant. Engineered pathogens change that calculus. Advances in synthetic biology make it increasingly feasible to design organisms that are simultaneously highly transmissible, highly lethal, and resistant to existing treatments.5PubMed Central. Existential Risk and Cost-Effective Biosecurity
The concern is not that a rogue actor could do this today, but that the tools are becoming cheaper, more widely available, and more powerful every year. Biological engineering offers extraordinary promise in medicine and materials science, but the same techniques that let researchers design viruses to target cancer cells could theoretically be turned toward designing pathogens optimized to evade immune systems.6Journal of Student Research. Existential Risk of Synthetic Biology: How Biological Engineering Can Help the World or Destroy It While the probability of a species-ending engineered pandemic in any given decade may be low, the expected harm is so vast that even small reductions in risk carry enormous value. This is one area where the gap between current governance and the pace of technological capability feels especially wide.
Artificial Intelligence and the Alignment Problem
AI has become the single largest contributor to many researchers’ extinction-risk estimates. Ord’s one-in-six figure, for instance, drew heavily on a one-in-ten probability he assigned to extinction from unaligned artificial intelligence alone.3PubMed Central. Extinction of the human species: What could cause it and how likely is it to occur? The core worry is straightforward in concept, even if the details are fiercely debated: if we build systems that are more capable than us at pursuing goals, and those goals are not perfectly aligned with human welfare, the consequences could be catastrophic. There is currently no reliable method for guaranteeing that a superintelligent system’s objectives stay aligned with ours.7arXiv.org. Artificial General Intelligence, Existential Risk, and Human Risk Perception
Beyond the alignment problem, researchers have identified additional pathways through which AI could pose existential risks: a competitive race between nations or corporations to develop powerful AI could lead to corners being cut on safety; AI could be weaponized in military or cyber applications that spiral out of control; and increasingly autonomous systems could make decisions that humans can no longer meaningfully oversee or reverse.8Oxford Handbook of Digital Ethics. How Does Artificial Intelligence Pose an Existential Risk? It is worth noting that reasonable experts disagree sharply about how likely any of these scenarios are. Some think the risk is wildly overstated, while others consider it the most urgent problem facing the species. The disagreement itself tells you something about how much genuine uncertainty exists.
Climate Change and the Limits of the Human Body
Unmitigated climate change would be devastating for civilization but is unlikely to cause human extinction on its own. The more immediate danger is regional uninhabitability. The human body cools itself through sweating, and that process stops working when the surrounding air is both extremely hot and extremely humid. For years, researchers cited a theoretical wet-bulb temperature of 35°C as the upper limit of human survivability, but laboratory testing of young, healthy subjects found the real threshold is meaningfully lower, averaging around 31°C, and dropping further in hot, dry conditions.9PubMed Central. Evaluating the 35°C wet-bulb temperature adaptability threshold for young, healthy subjects (PSU HEAT Project) Older adults, children, and people with health conditions would fare worse still.
Parts of South Asia, the Persian Gulf, and equatorial Africa already approach these limits during extreme heat events. Under high-emissions scenarios, these regions could see wet-bulb temperatures that regularly exceed human tolerance by late this century, making outdoor activity lethal without air conditioning. That would displace hundreds of millions of people and destabilize food systems, but it would not, by itself, render the entire planet uninhabitable. The extinction concern with climate change is less about heat directly and more about cascading failures: crop collapse, water shortages, mass migration, and the conflicts that follow. Climate change is better understood as a threat multiplier that makes every other risk on this list harder to manage.
Asteroids, Supervolcanoes, and Other Natural Wild Cards
These are the risks that capture public imagination but contribute relatively little to total extinction probability over the next few centuries. A truly massive asteroid impact, on the scale of the one that ended the age of dinosaurs, could render Earth uninhabitable for large animals. But such impacts are extraordinarily rare. For a civilization-ending strike by a giant asteroid, the probability over the next billion years is estimated between 0.03 and 0.3, and a giant comet impact in the next century comes in around two in a trillion.10Futures. Humanity extinction by asteroid impact We also have the beginnings of a planetary defense capability, with recent NASA missions demonstrating that asteroid deflection is at least technically possible.
Supervolcanic eruptions are a different story, because we have no way to prevent them and limited ability to predict them far in advance. The Toba eruption roughly 74,000 years ago was once thought to have nearly wiped out humanity, but more recent climate modeling suggests that super-eruptions may not produce the extreme cooling once assumed. Archaeological evidence does not clearly show increased human mortality following Toba.11Journal of Climate. Severe Global Cooling After Volcanic Super-Eruptions? The Answer Hinges on Unknown Aerosol Size This does not mean supervolcanoes are harmless; a major eruption today would cause severe disruption to agriculture and climate. But the evidence increasingly suggests they are not the species-enders they were once feared to be.
The Deep Future and the Brightening Sun
If humanity manages to navigate its self-created risks and the occasional cosmic dice roll, the ultimate deadline is set by astrophysics. The Sun grows brighter over geological time as it fuses hydrogen in its core. As it does, Earth’s surface temperature rises, and the planet’s carbon cycle draws more and more CO₂ out of the atmosphere to compensate. Eventually, atmospheric CO₂ drops below the level needed to sustain plant photosynthesis, and the terrestrial biosphere collapses. Most estimates place this endpoint at roughly a billion years from now, though recent modeling using updated climate parameters has pushed the estimate out to about 1.6 to 1.86 billion years.12The Astrophysical Journal Letters. Seafloor Weathering and Stochastic Outgassing Unlikely to Significantly Shorten the Future Lifespan of Earth’s Terrestrial Biosphere
Beyond that, the story gets even more dramatic. At some point, solar brightening will push enough energy into Earth’s climate to trigger significant water loss to space, as water vapor rises into the upper atmosphere and gets broken apart by ultraviolet radiation. Models suggest this process becomes rapid once the Sun’s output reaches about 21 percent above its current level, at which point global average surface temperatures would exceed 87°C.13Journal of Geophysical Research: Atmospheres. The evolution of habitable climates under the brightening Sun Long before that point, Earth would be uninhabitable for anything resembling current life. These are timescales on which the question shifts from “can humanity survive?” to “will our descendants have left the planet by then?”
Could Survivors Rebuild After a Catastrophe?
Even in the worst plausible near-term scenarios, total extinction is harder to achieve than you might think. Humans are spread across every continent and most ocean islands. Some fraction of the population would likely survive even a severe nuclear winter or global pandemic, especially in geographically isolated areas with independent food production. Researchers have specifically studied which island nations would serve as the best refuges against different types of global catastrophe.
For pandemic scenarios, Australia, New Zealand, and Iceland score highest on a composite index of resilience factors, a finding that the COVID-19 pandemic partially validated, since island nations generally performed better at border control.14PubMed. Optimizing Island Refuges against global Catastrophic and Existential Biological Threats: Priorities and Preparations For nuclear winter and other sunlight-blocking catastrophes, the same group of researchers found that Australia, New Zealand, Iceland, the Solomon Islands, and Vanuatu showed the greatest resilience, though even the best-positioned nations had serious vulnerabilities in energy supply, manufacturing, and trade dependencies.15PubMed. Island refuges for surviving nuclear winter and other abrupt sunlight-reducing catastrophes Islands in general offer natural advantages through geographic isolation and diverse conditions, and these advantages could be enhanced through deliberate preparation, such as underground bunkers and long-term data storage facilities.16foresight. Islands as refuges for surviving global catastrophes
Survival and recovery are two different things, though. Population genetics research indicates that a species needs an effective population of at least 500 to 1,000 individuals to remain genetically viable over the long term, depending on reproductive rate.17Biodiversity and Conservation. Prediction of the minimum effective size of a population viable in the long term Getting above that threshold is one challenge; maintaining the technological knowledge and social organization needed to prevent a permanent collapse of civilization is another. A few thousand survivors on a remote island could perpetuate the species biologically while losing most of what we think of as human civilization, perhaps permanently.
Geoengineering and the Trap of Dependency
One proposed response to climate change is solar geoengineering, which involves injecting aerosols into the upper atmosphere to reflect sunlight and cool the planet. If deployed at scale, it could mask significant warming. But this creates a dangerous dependency: if geoengineering were ever stopped suddenly, whether because of war, economic collapse, or political disagreement, temperatures would snap back upward rapidly. This rebound, known as termination shock, could be more damaging than gradual warming because ecosystems and human systems would have no time to adapt.18Earth’s Future. The Risk of Termination Shock From Solar Geoengineering
Termination shock is not itself an extinction risk, but it illustrates a broader pattern in how technological fixes can create new vulnerabilities. A civilization that becomes dependent on a single intervention to remain habitable is a civilization that has added a new single point of failure. The same logic applies to any technology that props up a large population beyond what the underlying environment can support: if the technology fails, the correction comes all at once.
Would Our Descendants Still Be “Human”?
There is a philosophical wrinkle in the extinction question that deserves attention. Even if a population bearing some relationship to us persists for millions of years, it is almost certain that evolution, genetic drift, and deliberate modification would change that population beyond recognition. The cumulative impact of genetic enhancement technologies could accelerate the transition to what would effectively be a new species.19PubMed. Genetic enhancement, human extinction, and the best interests of posthumanity This is not speculative hand-waving: the fossil record shows that no mammalian species persists unchanged for more than a few million years, and humans are already beginning to edit their own genome in ways that could compound over generations.
So the question “how long before humans go extinct?” has a strange possible answer: Homo sapiens as currently defined may not exist in a million years, not because we were wiped out but because we became something else. Whether that counts as “extinction” depends on how you define the term. From a biological standpoint, it absolutely does. From the perspective of our descendants, it might feel more like a graduation. The Doomsday argument, a probabilistic thought experiment formalized by philosopher Nick Bostrom, suggests that our position in the total sequence of all humans who will ever live gives reason to think the species will not persist as long as we might hope.20Think. The doomsday argument Whether you find that argument compelling or circular is a matter of some debate, but it does highlight that our intuitions about our species’ future are shaped by where we stand in its history.
The Off-World Question
Space colonization is frequently invoked as the ultimate insurance policy against extinction. If humans establish self-sustaining populations on other worlds, no single catastrophe confined to Earth could end the species. In practice, however, creating a truly self-sustaining colony is far harder than popular culture suggests. Existing experiments with closed ecological systems have demonstrated how difficult it is to maintain human life without ongoing inputs from Earth’s biosphere.21Futures. Biodiversity requirements for self-sustaining space colonies A Mars colony that depends on regular supply runs from Earth is not a hedge against extinction; it is an extension of the same fragile system.
The timeline for genuine off-world self-sufficiency, meaning a colony that could persist and grow even if all contact with Earth ceased, is anyone’s guess. Optimistic projections from space agencies and private companies tend to focus on getting people to Mars, not on the much harder problem of keeping them alive indefinitely once supply ships stop coming. A colony that meets the minimum viable population threshold and can produce its own food, medicine, energy, and manufactured goods from local resources is probably centuries away at a minimum. Until then, all of humanity’s eggs remain in one basket, and the urgency of managing Earth-bound risks stays high.