RNA is supposed to fall apart within hours, yet researchers have now read it from a mammoth frozen in Siberian permafrost 39,000 years ago — and the transcripts caught his muscle tissue in the middle of a stress response

RNA is the molecule biologists generally treat as disposable. Unlike DNA, which is built to be a stable long-term archive, RNA is manufactured, used, and broken down by a cell within hours, sometimes minutes, as part of ordinary business: turning genetic instructions into proteins, then discarding the messenger once the job is done. Outside a living cell, without anything actively protecting it, most RNA is expected to degrade within hours to days. That is the baseline assumption researchers have worked from for decades.

A team led by Emilio Mármol Sánchez, now at the University of Copenhagen’s Globe Institute, working with Love Dalén at Stockholm University’s Centre for Palaeogenetics and colleagues including Marc Friedländer and Bastian Fromm, has now recovered and sequenced RNA from muscle tissue belonging to a woolly mammoth frozen in Siberian permafrost for roughly 39,000 years. The work was published in Cell in November 2025.

This is one study built around a specific set of specimens, not a settled claim about how long RNA can survive under any conditions. Permafrost offers something close to ideal preservation: sustained sub-zero temperatures that slow the chemical and microbial processes that normally break RNA down. What the paper shows is that under those particular conditions, over that particular span of time, enough RNA survived intact to be read.

Yuka’s last stress signal

The most complete results came from Yuka, a juvenile woolly mammoth recovered from Siberian permafrost and one of ten permafrost-preserved mammoths sampled for the study. Genetic analysis in this study revised Yuka’s sex to male, correcting earlier field assessments.

The woolly mammoth genome carries more than 20,000 protein-coding genes, and from Yuka’s frozen leg muscle the team recovered several hundred active transcripts, a small fraction of that total but a striking amount to find intact at all in tissue this old. Among them were genes associated with muscle contraction and metabolic regulation under physiological stress, the kind of activity a cell ramps up when an animal’s body is under acute strain. Mármol described the significance plainly in comments carried by EurekAlert: “We found signs of cell stress, which is perhaps not surprising since previous research suggested that Yuka was attacked by cave lions shortly before his death.”

That earlier finding, that Yuka bore wounds consistent with a cave lion attack, came from separate prior research on the specimen’s remains. What the new RNA data adds is a molecular echo of the same event: evidence that Yuka’s muscle cells were actively responding to stress at or near the point of death, rather than simply a body that happened to carry old injuries. Dalén, speaking to NPR, put it this way: “You’re actually seeing processes going on inside the cells right around the time it died. And these processes have then been frozen in time for 40,000 years.”

Proving the signal is real, not contamination

Ancient biomolecule research has a persistent contamination problem. Any sample handled by modern researchers, stored in modern facilities, or exposed to modern environmental microbes risks picking up genetic material that has nothing to do with the specimen itself. For DNA, researchers have developed decades of authentication techniques to separate genuine ancient signal from contamination. RNA authentication is a newer, less established discipline, which makes the question of proof unusually important here.

The team leaned on microRNAs, short regulatory RNA molecules, as their authentication tool: rare mutations within specific microRNA sequences matched known mammoth genetic markers rather than anything from a modern contaminant. Fromm called this, in comments carried in the same EurekAlert release, “a smoking-gun demonstration of their mammoth origin,” while Friedländer described the muscle-specific microRNAs as “direct evidence of gene regulation happening in real time in ancient times.” The distinction matters: without that authentication step, a stress-response signal recovered from 39,000-year-old tissue would be a much harder claim to stand behind.

What this does and doesn’t open up

It’s tempting to read a result like this as an opening onto a new field of ancient physiology, reconstructing the last moments of long-dead animals at scale. This is a single study, working from a small number of well-preserved specimens under unusually favourable freezing conditions, and permafrost mammoths are a poor model for what’s recoverable from remains preserved anywhere else on Earth. Warmer, wetter, or more disturbed burial environments degrade RNA far faster, and nothing in this paper suggests those limits have moved.

What the study does establish is a lower bound: RNA can survive for tens of thousands of years, under the right conditions, in a form complete enough to say something specific about what an animal’s cells were doing shortly before it died. Whether that finding extends to other frozen specimens, other tissue types, or other timescales is a question for research that hasn’t been done yet.