The deadliest mountain in the world isn't the one you think
Everest has the most deaths, not the worst odds. We ranked the world's deadliest mountains by death rate using the Himalayan Database's newest data.

On this page (16)
- Death rate vs death toll, the difference that changes the answer
- The deadliest mountains in the world, ranked
- 1. Annapurna I, the worst odds of any 8,000er
- 2. Dhaulagiri I, the deadliest of the modern era
- 3. K2, the savage mountain
- 4. Nanga Parbat, the Killer Mountain
- 5. Kangchenjunga and Makalu, remote and steep
- 6. Everest, the highest toll but not the highest rate
- 7. Manaslu, popular and avalanche-prone
- 8. Denali, cold as a weapon
- 9. The Matterhorn and the Eiger, the Alps’ deadliest
- Why the death rates keep falling
- You don’t have to climb them, you can stand beneath them
- And if you truly want to climb, start small and guided
- Frequently asked questions
- Keep reading
By death rate, the deadliest mountain in the world is Annapurna I in Nepal: 75 people have died on it against 557 successful summits, about 13.5 deaths for every 100 summits, according to the Himalayan Database. By death toll, it is Everest, with 344 deaths, simply because far more people climb it. And if you only count climbs since 2000, the worst odds belong to a third mountain, Dhaulagiri I.
We pulled those numbers ourselves from the Himalayan Database’s newest data file (the Autumn-Winter 2025 update, downloaded on 4 October 2026), so they are fresher than the 30%-plus figures still repeated around the web. Below is the full ranking, why each mountain kills, and why trekking to the foot of these peaks is a different world from climbing them.
- Deadliest by rate: Annapurna I, 13.5 deaths per 100 summits all time.
- Deadliest by toll: Everest, 344 deaths, but only 2.5 per 100 summits.
- Deadliest since 2000: Dhaulagiri I at 9.3 per 100; Annapurna has fallen to 4.5.
- Trekkers are not climbers: Everest Base Camp is 3.5 vertical kilometres below the summit, and the real trekking risk is altitude sickness.
Death rate vs death toll, the difference that changes the answer
“Deadliest” means two different things, and they give two different winners. The death toll is how many people have died on a mountain. The death rate is how many die for every successful summit, which is the closest public measure of how likely a climber is to die there.

Everest leads the toll because it has more than 13,000 recorded summits; its rate is low. Annapurna I has the worst rate because it kills often and is summited rarely. A climber on Annapurna has historically faced roughly five times Everest’s odds per summit.
Both numbers below come from the Himalayan Database, the non-profit archive built on the late Elizabeth Hawley’s records, which covers every recorded expedition to Nepal’s major peaks from 1905. Deaths include hired Sherpa and other staff, who make up a large share: 134 of Everest’s 344. The database does not cover K2 or Nanga Parbat, which are in Pakistan, so those two use other counts and are marked as such.
The deadliest mountains in the world, ranked
Annapurna I tops the ranking by rate, with Dhaulagiri I close behind and K2 third; Everest is fifth among Nepal’s giants by rate but first by toll. Here are the numbers, with the date each one runs to.
| Mountain | Height | Deaths | Summits | Deaths per 100 summits | Data to |
|---|---|---|---|---|---|
| Annapurna I (Nepal) | 8,091 m | 75 | 557 | 13.5 (4.5 since 2000) | Winter 2025, HDB |
| Dhaulagiri I (Nepal) | 8,167 m | 92 | 739 | 12.4 (9.3 since 2000) | Winter 2025, HDB |
| K2 (Pakistan/China) | 8,611 m | 96 | about 800 | about 12 | Aug 2023, estimate |
| Kangchenjunga (Nepal/India) | 8,586 m | 54 | 682 | 7.9 (2.6 since 2000) | Winter 2025, HDB |
| Makalu (Nepal/China) | 8,485 m | 53 | 890 | 6.0 (4.4 since 2000) | Winter 2025, HDB |
| Everest (Nepal/China) | 8,849 m | 344 | 13,565 | 2.5 (1.4 since 2000) | Winter 2025, HDB |
| Manaslu (Nepal) | 8,163 m | 90 | 3,898 | 2.3 (1.0 since 2000) | Winter 2025, HDB |
| Nanga Parbat (Pakistan) | 8,126 m | 31 before the first ascent | — | no current count we trust | 1953 |
| Matterhorn (Switzerland/Italy) | 4,478 m | about 500 | — | — | 1865–2011 |
| Eiger North Face (Switzerland) | 3,967 m | at least 64 | — | — | since 1935 |
For the full set of Nepal’s 8,000-metre peaks, Lhotse sits at 1.7 deaths per 100 summits and Cho Oyu at 1.3, the safest of them. Every rate here is deaths divided by summits; a person who summits twice on separate expeditions counts twice, as the database records it.
1. Annapurna I, the worst odds of any 8,000er
Annapurna I has recorded 75 deaths against 557 summits, a rate of 13.5 per 100, the highest of Nepal’s eight 8,000-metre peaks. Guinness World Records lists it as the deadliest Himalayan mountain on the same database (13.42% as of 7 February 2025; the small difference is the 2025 seasons).

Avalanches are the killer. In the database, 39 of the 75 deaths are coded as avalanche and 18 as falls, and 43 happened while teams were still preparing the route, before any summit bid. Guinness notes that most deaths occur at a single avalanche point at about 5,900 m between Camp II and Camp III. It was the first 8,000er ever climbed, by Maurice Herzog’s French team in 1950, and has never been tamed. The rate has fallen, to 4.5 per 100 since 2000, because more people now reach the top, not because the mountain changed.
The trek to its base camp is a different thing entirely, more on that below. We dig into the mountain itself in why Annapurna I is the world’s deadliest mountain.
2. Dhaulagiri I, the deadliest of the modern era
Dhaulagiri I has 92 deaths against 739 summits, 12.4 per 100, and since 2000 it has the worst rate of any Nepal 8,000er: 41 deaths against 441 summits, 9.3 per 100. That is the number most lists miss, because they only quote all-time figures.
Avalanches (41 deaths) and falls (27) dominate. The mountain rises almost alone above the deep Kali Gandaki gorge, so there is little shelter from wind, and its remoteness keeps rescue slow.
3. K2, the savage mountain
K2 has an estimated 800 summits and 96 deaths as of August 2023, about 12 deaths per 100 summits. It is in Pakistan and China, so the Himalayan Database does not cover it, and its count is a compiled estimate rather than a database tally. The rate used to be far higher; a rise in successful ascents in recent seasons has pulled it down.
Everything on K2 is steep and exposed, the final push passes under the hanging ice cliff of the Bottleneck, and above 8,000 m rescue is effectively impossible. For the Karakoram peaks side by side, see the Karakoram death-rate explainer.
4. Nanga Parbat, the Killer Mountain
Nanga Parbat earned its nickname before anyone stood on top: 31 people had died on it by the time Hermann Buhl made the first ascent in 1953. In 1937 a single avalanche killed sixteen men, seven Germans and nine Sherpas.
Most lists give it a modern rate of around 20%, but we could not find a current, independently compiled count of its deaths and summits that we trust, so we leave it out of the rate column rather than print a guess. Its Rupal Face is one of the biggest mountain walls on Earth.
5. Kangchenjunga and Makalu, remote and steep
Kangchenjunga, the world’s third-highest mountain, has 54 deaths against 682 summits (7.9 per 100); Makalu, the fifth-highest, has 53 against 890 (6.0). Both are far from roads and help. Since 2000, Kangchenjunga has dropped to 2.6 per 100 while Makalu has stayed at 4.4, the third-worst modern rate in Nepal after Dhaulagiri and Annapurna.
6. Everest, the highest toll but not the highest rate
Everest has recorded 344 deaths against 13,565 summits, 2.5 per 100, and 1.4 per 100 since 2000. Its toll dwarfs every other mountain because so many people try it.
The database shows where Everest kills. Avalanches account for 78 deaths, falls 75 and altitude sickness 49. And the way down is worse than the way up: 126 deaths happened descending from a summit bid against 35 on the ascent, when climbers are exhausted and short of oxygen and daylight (more in why more climbers die on the descent). Another 128 died while routes were being prepared, mostly hired staff in the Khumbu Icefall. Its worst years in the database are 2023 (18 deaths), 2014 (17) and 2015 (14); 2025 had four deaths against 850 summits. The 2026 season, not yet in the database, produced five deaths against a record 1,008 summits.
7. Manaslu, popular and avalanche-prone
Manaslu has 90 deaths against 3,898 summits, 2.3 per 100, and just 1.0 since 2000. Older lists that put it near 10% are out of date. It has become a popular “first 8,000er”, and its worst year in the database, 2012, recorded 13 deaths. The Manaslu Circuit trek loops around it at walking altitudes and shares almost none of that risk.
8. Denali, cold as a weapon
North America’s highest peak, 6,190 m, killed 96 people from 1903 to 2006 according to a study in High Altitude Medicine & Biology, a fatality rate of 3.08 per 1,000 summit attempts, and five more died on its West Buttress route in 2026. One of them was a National Park Service climbing ranger. The park service, which now calls the mountain Mount McKinley, registered 976 climbers that season, and 582 reached the summit. Its sub-Arctic latitude makes it colder and stormier than its height suggests.
9. The Matterhorn and the Eiger, the Alps’ deadliest
Zermatt’s tourism office estimates about 500 climbers died on the Matterhorn between its first ascent in 1865 and 2011, and at least 64 climbers have died on the Eiger’s north face since 1935. The Matterhorn’s first ascent ended with four of the seven-man party falling to their deaths on the descent. Today some 3,500 people a year try it without a guide, on loose rock where one slip is enough. The Eiger’s Nordwand, 1,600 to 1,800 m of rock and ice, earned the name Mordwand, the murder wall.
Why the death rates keep falling
Mostly because the bottom of the fraction grew: commercial seasons added thousands of summits while deaths rose slowly. Annapurna’s all-time rate is 13.5 per 100, but only 4.5 since 2000. Everest went from 14.8 deaths per 100 summits before 2000 to 1.4 since.
Better forecasting, fixed ropes, bottled oxygen and experienced Sherpa teams helped too. But a rate can also drop when a mountain gets busier without getting safer, which is why the since-2000 column matters. Dhaulagiri’s 9.3 says the danger there is still very current.
You don’t have to climb them, you can stand beneath them
Almost none of the danger on this page applies to trekking. The deaths above happen in mountaineering: roped climbing, fixed lines, crampons and the death zone above 8,000 m. Trekking to a base camp is a long walk on established trails, sleeping in teahouse lodges, thousands of metres below where the killing happens.
- Everest Base Camp sits at 5,364 m, about 3.5 vertical kilometres below the summit. You never touch a rope or the Khumbu Icefall. See our 14-day Everest Base Camp Trek (US$1,850 solo, less each in a small group) and the free Everest Base Camp trek guide.
- Annapurna Base Camp sits at 4,130 m in the amphitheatre below the deadliest mountain on this list, with none of its avalanche slopes. See our 6-day Annapurna Base Camp Trek (US$695 solo) and the Annapurna Base Camp trek guide.
The honest risk on these treks is altitude sickness, which is real and largely manageable with a sensible ascent profile, rest days and a guide who will turn you around. We cover it in our complete guide to altitude sickness in Nepal, and how hard Everest Base Camp really is gives the physical picture. In more than 5,000 treks since 1998 we have had zero client fatalities.
Want to see these giants up close, safely? Tell us your dates and fitness and our guides will suggest the right base camp trek. Plan my trek
And if you truly want to climb, start small and guided
Do not start on an 8,000er; start on a trekking peak with a qualified guide. These let a fit beginner learn crampons, ropes and altitude with real margins:
- Island Peak (Imja Tse), 18 days, a classic first Himalayan summit at 6,189 m, US$3,600 solo.
- Mera Peak, 15 days, Nepal’s highest trekking peak at 6,476 m, non-technical but genuinely high, US$2,500 solo.
Frequently asked questions
What is the deadliest mountain in the world?
By death rate, Annapurna I in Nepal: 75 deaths against 557 summits, about 13.5 per 100, in the Himalayan Database through the Autumn-Winter 2025 update. By total deaths, Everest, with 344. Since 2000, Dhaulagiri I has the highest rate at 9.3 per 100 summits.
Which mountain has the most deaths?
Everest, with 344 deaths recorded through the 2025 seasons, plus five more in spring 2026. It has the most deaths because it has by far the most climbers, more than 13,500 summits.
Is K2 deadlier than Everest?
Yes, by rate. K2 has about 12 deaths per 100 summits (96 deaths, about 800 summits, as of August 2023) against Everest’s 2.5. Everest has more deaths in total.
Why is Annapurna so dangerous?
Avalanches. More than half its 75 recorded deaths are coded as avalanche, and Guinness says most happen at one avalanche point between Camp II and Camp III. They strike whether you are climbing well or not. Few people summit it, which keeps its rate high.
What is the safest 8,000-metre peak in Nepal?
Cho Oyu, with 52 deaths against 4,124 summits, about 1.3 per 100, followed by Lhotse at 1.7. Safest still means a real risk of dying.
Is trekking to Everest Base Camp dangerous?
Far less than climbing. Base Camp is at 5,364 m, about 3.5 km below the summit, reached on a walking trail with no technical climbing. The main risk is altitude sickness, which a steady pace and acclimatisation days largely manage.
Keep reading
The avalanche slope behind the world’s worst odds.
Why the walk down kills more than the climb up.
Five deaths against a record 1,008 summits.
The other side of these peaks.
Curious about the country these giants call home? Start with Nepal Himalaya facts, our best treks in the world, and the Nepal trekking permits hub when you are ready to plan.
Sources: The Himalayan Database, HIMDATA file with the Autumn-Winter 2025 update (himalayandatabase.com), downloaded and tallied by Travel Himalaya Nepal on 4 October 2026, with cause and phase of death from the database’s own codes; Guinness World Records, Deadliest Himalayan mountain; Wikipedia, K2 and Nanga Parbat (K2 count as of August 2023); US National Park Service, 2026 Denali season wrap-up; McIntosh et al., Mountaineering fatalities on Denali, High Altitude Medicine & Biology (2008); Zermatt Tourism, Matterhorn; Wikipedia, Eiger. All read 3–4 October 2026.
Photos via Pexels: Grisha Grishkoff (cover), ashim bhatta (Annapurna I).
Ready to Trek?
From reading about it to standing on it
Our Pokhara-based guides have been doing this since 1998. Tell us your dates and fitness level — we'll build your perfect itinerary. Free, no obligation.
More guides like this
Popular trips to consider
View all treks & toursFree Trekker’s Insider Guide
Permits, packing lists, cost breakdowns — no fluff.
We send one useful email. You can unsubscribe anytime.

