Oxygen in the air we breathe may play a significant role in triggering lung cancer, new research suggests.
Scientists in the US found that rates of the disease decreased dramatically at higher altitudes, where the air is thinner.
Incidence fell by 7.23 cases per 100,000 individuals for every 1,000 metre (3,281 feet) rise in elevation.
An estimated tens of thousands of cases of lung cancer in the US would be avoided if the whole population lived around 3,000 metres (9,800 feet) above sea level, said the researchers.
The results suggested "substantial evidence" for an inhaled cancer-trigger "tied directly to elevation".
Writing in the journal PeerJ, the two American authors concluded: "Viewing our findings through the lens of the literature, atmospheric oxygen emerges as the most probable culprit."
Although oxygen is essential for life, it is also known to be highly reactive and potentially carcinogenic.
When cells in the body use oxygen to harness energy stored in food, they produce a dangerous natural by-product, oxygen free radicals.
These destructive molecules, also known as reactive oxygen species (ROS), can inflict serious damage on cell structures and DNA.
Kamen Simeonov, from the University of Pennsylvania, and Daniel Himmelstein, from the University of California at San Francisco, compared cancer rates across 250 western counties in the US with varying altitude levels.
They found a strong association between elevation and lung cancer - but not breast, prostate and bowel cancer - indicating a role played by something inhaled.
When other risk factors were taken into account, the influence of altitude sharpened for lung cancer and showed a "remarkably strong effect size and significance".
Exposure to sunlight and pollution, both of which are affected by elevation, had no impact.
Last year, lung cancer killed an estimated 159,260 people in the US.
The scientists wrote: "Were the entire United States situated at the elevation of San Juan County (3,473 metres), we estimate 65,496 fewer lung cancer cases would arise per year."
They concluded: "Overall, our findings suggest the presence of an inhaled carcinogen inherently and inversely tied to elevation, offering epidemiological support for oxygen-driven tumorigenesis."