From 313 ppm to 426: the number Charles Keeling started
When Charles David Keeling took his first reading on Mauna Loa in Hawaii on 29 March 1958, he got 313 parts per million of carbon dioxide16. In 2025 the air around the planet averaged 425.6 ppm1, and it is still rising by about two and a half ppm a year10.
Parts per million counts molecules: out of every million molecules of dry air, about 426 are now CO2. That’s only about 0.04% of the air, yet each extra part per million is roughly 7.8 billion tonnes of CO210. The level has gone up in every year of the global record. Even the slowest year, 1964, added 0.49 ppm3.
Why you may see a different number elsewhere
Search for today’s CO2 level and you’ll find figures anywhere from about 424 to 432 ppm. Most of them are right; they measure different things.
The figure on this page is the global average, taken from marine surface sampling sites around the world1. It’s the same kind of global figure the World Meteorological Organization (WMO) reports, which put 2024 at 423.9 ppm4. Mauna Loa, the station most news reports quote, is a single site, and in 2025 it averaged 427.35 ppm2 against the global 425.621. The number at the top of this page runs a little higher than both because it carries the 2026 reading forward along a projection, explained under “How this number was made”.
Mauna Loa also swings through the year. Keeling was the first to notice why: in the Northern Hemisphere CO2 peaks in May, falls as plants grow through the summer, and rises again as they die back in autumn16. In 2025 that swing ran from 430.51 ppm in May to 424.37 in September2. The headline figures, like 432.3 ppm in May 202616, are that yearly peak.
278 ppm before the factories
Before the industrial revolution, the air held about 278 ppm of CO27. Ice cores, which hold bubbles of ancient air, show it never went above 300 ppm in the 800,000 years before that13. Over all that time CO2 rose and fell with the ice ages, swinging by 50 to 100 ppm between the depths of an ice age and the warm spells in between7.
Since 1750 we have added about 147 ppm. That is more than the whole difference between an ice age and a warm period, added in under three centuries. The rise of the last 60 years has been 100 to 200 times faster than the natural increase at the end of the last ice age13.
That's how much more carbon dioxide the air held in 2025 than before 1750
CO2 was last this high more than two million years ago
Carbon dioxide has not been this high for at least two million years, and probably not for about 14 million. The first figure is the IPCC’s, stated with high confidence6. During a warm spell in the mid-Pliocene, about 3.2 million years ago, CO2 sat between 360 and 420 ppm7. We have now passed the top of that range.
The second figure comes from a 2023 study that pieced together 66 million years of evidence. It puts the last time average CO2 was as high as today at about 14 to 14.5 million years ago12. You may also see “16 million years” quoted. That comes from the same study but answers a different question: when CO2 was last consistently higher than today12.
Rising 2.6 ppm a year, because half of what we emit stays up
CO2 is rising by about 2.6 ppm a year, the average of the last decade10. That’s roughly 20 billion tonnes of CO2 added to the air every year. In the 1960s it rose by 0.8 ppm a year4. In 2024 it jumped 3.5 ppm, the biggest one-year rise since modern measurements began4, and in 2025 it fell back to about 2.110.
Those swings come from what the land and oceans take back, since emissions change too slowly to cause them. Over the past decade, about half of all the CO2 humans released stayed in the air. The oceans absorbed 29% and plants and soils on land took 21%10. Scientists call these carbon sinks. In a hot, dry El Niño year like 2024, fires burn more and the sinks take up less, so more of each year’s emissions stays up4.
Warming is also wearing the sinks down. Between 2015 and 2024, climate change cut the land sink by 25% and the ocean sink by 7.1%10. A 2025 study found that this weakening has added 8.3 ppm to the air since 1960, and that heat and deforestation have turned forests in Southeast Asia and large parts of South America from absorbing carbon to releasing it11.
When the pandemic cut emissions in 2020, the drop didn’t stand out from the natural ups and downs of the CO2 record14. It takes years of lower emissions, not months, before the level responds6.
CO2 has a speed limit, not a safe level
The 1.5 °C target in the Paris Agreement is a limit on warming, not on CO2, so there’s no official ppm line to measure against. It does imply a speed, though.
The IPCC’s 1.5 °C pathways need CO2 to rise by only 1.33 to 1.79 ppm a year on average through the 2020s, and in the 2030s to almost stop rising, or in one pathway to start falling. From 2020 to 2025 Mauna Loa recorded 2.61 ppm a year15, roughly half again as fast as the fastest pace a 1.5 °C path allows. Every year spent above that pace leaves a steeper cut for the years after.
507 ppm by 2050 on today’s policies
The IPCC’s middle scenario (SSP2-4.5) warms the planet by about 2.7 °C6. That’s close to what the UN Environment Programme finds for current policies (up to 2.8 °C)8 and what Climate Action Tracker finds (2.6 °C)9, which makes it the nearest stand-in for where today’s policies lead. On that path, CO2 reaches about 507 ppm by 2050 and 603 ppm by 21005. The air of 2100 would hold more than twice the CO2 it held before 1750.
The other scenarios spread widely. By 2050 they range from 438 ppm, in the one that holds warming near 1.5 °C, to 563 ppm on the highest-emission path. By 2100 the spread is 394 to 1,135 ppm5.
That CO2 is what drives the warming, and the temperature line follows the same scenario.
CO2’s rise is slowing, but it is still climbing
Once the Pacific’s natural warm and cool swings (El Niño and La Niña) are taken out, the yearly rise in CO2 has been growing more slowly since about 2012, because global emissions have grown more slowly15. Between 2015 and 2024, 35 countries cut their fossil CO2 while their GDP grew, nearly double the 18 of the decade before, and emissions from clearing land have been falling since the late 1990s10.
None of that has stopped the level rising. If the last decade’s pace of about 2.6 ppm a year simply held, a rough extrapolation rather than a forecast, the air would pass 450 ppm around 2035 and 500 ppm in the mid-2050s. What could speed that up is the sinks. Forests that already release more carbon than they absorb11 are worth watching, because every sink that weakens leaves more of each year’s emissions in the air.
The lowest scenario shows the line can turn, though. In it, CO2 peaks and falls back to 394 ppm by 21005, below where it stands today. The IPCC finds that deep, sustained emission cuts would show up in the concentration within years6.
What goes into the air each year, and how fast it’s still growing.
