How did life begin?

Earth formed about 4.54 billion years ago. Fossils generally treated as undisputed put life on Earth by 3.35 billion years ago, and by 3.47 on some counts. Carbon in Greenland rocks more than 3.7 billion years old has been read as biological, though that reading is still argued over, and contested but serious candidates push back further still: carbon isotope signatures in a 4.1-billion-year-old zircon, and structures in hydrothermal vent precipitates at least 3.77 and possibly 4.28 billion years old. A 2024 molecular-clock study puts the last common ancestor of all cellular life at about 4.2 billion years ago, which would put the origin earlier again.

Whatever happened, happened fast in geological terms. Life did not take billions of years to get started. Liquid water was on the surface by about 4.4 billion years ago, and life was there within a billion years of it, perhaps within a few hundred million.

In 1953 Stanley Miller ran electricity through a mixture of water, methane, ammonia and hydrogen, and produced amino acids. The experiment has been refined and criticised for seventy years, not least because an atmosphere rich in methane and ammonia would not have lasted long. The finding that survives is narrower than its reputation: some of the small molecules life is made of form readily from simple ingredients, though often in low yields and among many other products.

Amino acids, nucleobases and sugars have all since been found in meteorites, and in 2025 amino acids and all five nucleobases of DNA and RNA turned up in pristine samples returned from the asteroid Bennu. Interstellar clouds carry less than they are usually said to: the simplest sugar was identified in one in 2000, and a reported detection of an amino acid in three of them was examined line by line in 2005 and did not hold. The raw material of biology is still not rare. It is, as far as we can tell, ordinary cosmic chemistry.

Every organism that has ever been sequenced uses nucleic acids to store information, proteins to do work, and a nearly identical genetic code to translate between them. The departures that have been found are small: reassigned codons in more than thirty lineages, among them mitochondria, some ciliates, green algae and yeasts, and a few groups of bacteria including Mycoplasma.

A formal test of common ancestry, run on ubiquitously conserved proteins against models in which major groups arose independently, came out as what its author called “powerful statistical evidence corroborating the monophyly of all known life”. Critics replied that the test measured how similar the proteins are rather than whether they share an ancestor; its author answered that the nested pattern of the similarities, not their degree, is what decides it. Neither side doubts the wider evidence for common ancestry, which the critics call overwhelming. None of that is evidence that life only started once. It is evidence that all known life descends from one lineage, whatever else may have started and died out.

Several settings have been proposed, and the argument is sharpest between two of them. Alkaline hydrothermal vents offer mineral catalysts, protected pores and, on their advocates’ model, natural proton gradients: a ready-made version of the energy machinery cells still use. Surface pools subject to wet–dry cycling and ultraviolet light offer chemistry that has worked in the laboratory, including an efficient synthesis of RNA building blocks.

Both scenarios have serious advocates and real evidence. Neither has produced a living system. The disagreement is not decorative; it goes to the question of whether life began in the dark or in the light.

The RNA world hypothesis proposes that RNA, which can both carry information and catalyse reactions, preceded both DNA and proteins. It has shaped the field for decades, and one review finds strong evidence that an RNA world did exist before DNA and proteins, while calling the case that life began with RNA more tenuous. It also requires RNA to assemble and copy itself under prebiotic conditions, which has never been demonstrated end to end. In 2025 an RNA polymerase ribozyme copied RNA strands, including a fragment of itself, but not the whole of itself.

Metabolism-first models invert the order: self-sustaining chemical cycles come first, and inheritance is bolted on later. They face the opposite problem. It is unclear how such a cycle could evolve without something to carry information between generations.

There is no accepted account of how a mixture of organic molecules becomes a system that copies itself, makes errors, and is therefore subject to natural selection. Every proposal has a gap in the middle, and the gaps are in different places.

This is the actual unknown. Not the ingredients, not the timing, not the setting: the step where chemistry starts keeping records.

We have one example of life, and one sample is a poor basis for calculating odds. If life emerges readily wherever conditions allow, the universe should be full of it. If the transition required an extraordinary coincidence, we may be alone. The one clue is how early life appeared, and how much it weighs is argued over. One Bayesian analysis finds the early start consistent with life being arbitrarily rare. Another, updated in 2025 with the 4.2-billion-year date for the common ancestor, puts the odds of a quick origin on Earth-like planets at better than ten to one, while granting that such planets may themselves be rare. Neither settles which picture is true.

A second, independent origin of life on Earth (a “shadow biosphere” with different chemistry) has never been found. The three tools that survey the microbial world, microscopy, cultivation and amplification of rRNA genes, are each built around the biochemistry of the only life we know, and the argument is that none of them would register an organism outside it. On that argument, not finding one is not evidence of its absence. Its proponents wrote in 2009 that the idea had had very little attention compared with the same question asked about other planets.

Alkaline hydrothermal vents, their advocates argue, would have held continuous proton gradients across the mineral walls of their pores, with alkaline vent fluid on one side and a more acidic ocean on the other. Modern cells harvest energy from proton gradients too. On this view, life did not invent its energy system. It inherited one from a rock and slowly took over the machinery. Critics answer that those walls are about a thousand times thicker than a cell membrane, which would leave almost no usable energy.

Repeated wetting and drying concentrates molecules and drives them to bond. Ultraviolet light, a problem for delicate biochemistry, also drives the synthesis that produced precursors of RNA, protein and lipids from the same starting chemistry, though in steps that need different reagents delivered in turn rather than all at once. Its advocates count its laboratory record the stronger; vent advocates count its continuity with how cells work today the weaker.

Panspermia proposes that life, or its precursors, arrived from elsewhere. Organic molecules certainly arrive on meteorites, and bacterial spores shielded from sunlight have survived up to six years in space. But relocating the question does not answer it: life would still have had to begin somewhere.