Rasuwa flood, live

What happened

At about 09:00 Nepal time on 26 August 2026, the Bhote Koshi rose suddenly where it crosses into Nepal at Rasuwagadhi. It was not raining in the Rasuwa catchment. Within minutes the water was through Timure, the nearest Nepali settlement, taking nine bank branches and the customs post with it. Syabrubesi, downstream and the gateway to the Langtang trek, lost its helipad, which slowed rescue flights into the upper valley on the first day, when they mattered most. Around a dozen riverside settlements in northern Rasuwa were hit in that first hour.

The flood did not stop at the district line. The Bhote Koshi joins the Trishuli at Syabrubesi, and the Trishuli runs on into the Narayani. The surge moved through Betrawati and Trishuli Bazar in Nuwakot, Galchhi and Dhunge Bazaar in Dhading, and past Muglin. Bodies have since been recovered as far as Chitwan and Nawalparasi. The Armed Police Force alone recovered 121 from the Narayani by Thursday afternoon.

Counts rose sharply and repeatedly through the first two days as search teams reached areas that had been cut off. The Prime Minister’s Office reported 95 dead on Wednesday evening and NDRRMA 72; by 12 October 2026, 6:00 am NPT Nepal Police reported 1456, and by 14 October 2026, 12:00 pm NPT Nepal Police reported 1458. Every figure on this page is provisional for that reason.

See the full hour-by-hour timeline.

The route the water took

Open the interactive map of the flood path, with the coordinates for each point.

What may have caused it

No government has confirmed the cause. Ordinary rainfall is ruled out. It was not raining in the catchment that morning. These are the explanations under examination, ranked by how strongly they are currently supported:

1. An ice-and-rock avalanche dammed the Lhende Khola, and the dam then burst (Leading explanation)
   Nepal’s Department of Hydrology and Meteorology and independent glaciologists, working from satellite imagery shared by Chinese counterparts through ICIMOD, put forward a preliminary hypothesis: an ice-and-rock avalanche blocked the Lhende Khola, a Bhote Koshi tributary on the Nepal-China border, roughly 20 km upstream of the Miteri Bridge. That formed a temporary debris-dammed lake, which then burst. This is described as preliminary, and no government has confirmed it.
2. An ice and rock fall off Langtang Lirung may have generated the flood directly, without a lake bursting (Newly proposed)
   Santosh Nepal, a water resources and climate change researcher, told Onlinekhabar on Friday that satellite images point to a large mass of ice, snow and rock falling from about 5,000 to 5,500 metres on Langtang Lirung’s northern slope and dropping nearly 2,000 metres into the Lhende Khola. He said the impact’s energy may have rapidly melted snow and ice, sending a sudden surge of water downstream without a lake first forming and bursting. He compared the mechanism to Uttarakhand’s 2021 Chamoli disaster, and said it is too early to conclude what happened without more satellite data and a field investigation. This is one researcher’s preliminary reading, not a government or ICIMOD finding.
3. The 08:37 seismic signal was a glacial collapse, not an earthquake (Reassessed)
   The US Geological Survey initially logged a magnitude 4.4 earthquake at 08:37 local time, minutes before the flood. It has since revised that analysis and concluded the signal was in fact a magnitude 5.2 glacial collapse rather than a tectonic earthquake, based on nearby seismic stations, long-period seismic waves and satellite imagery. In other words the tremor was not a separate trigger; it was the collapse itself being recorded.
4. Glacial lake drainage on the same river before (Related hazard)
   A broadly similar flood on the same river in July 2025 was ultimately attributed to the drainage of a supraglacial lake, meltwater pooled on the surface of a glacier. ICIMOD scientists say cascading glacial hazards of this kind are increasing at an "unprecedented" pace across the Hindu Kush Himalaya.

Read the full explanation of what caused the flood, including what a GLOF is and why the distinction matters for the districts downstream.

Damage

The Nepal Electricity Authority reports 14 hydropower projects damaged along the corridor, with a combined capacity of about 748 MW: 9 operating plants (354 MW) and 5 under construction (394 MW). The Department of Roads confirms the entire 42 km road from Betrawati to the Rasuwagadhi crossing was destroyed at multiple points, including several concrete bridges, cutting Nepal’s main overland trade and pilgrimage route to Tibet.

The hydropower damage in detail, project by project, and what it means for supply.

Physical Infrastructure Minister Sunil Lamsal put road and bridge damage alone at about Rs 200 billion on 26 August, and called the figure preliminary. Hydropower and private property are not in it.

See the full damage assessment, project by project.

Who is missing, and what that number means

Nepal Police reported 5285 people out of contact in its 3 October 2026, 12:00 pm NPT update. That figure is a contact list, not a casualty list: phone lines, power and internet are down across the valley and are coming back slowly, so being out of contact does not mean dead.

Most of the missing foreign travellers were moving toward the Kailash Mansarovar pilgrimage route through Tibet. Chinese state media separately report 546 people missing around Gyirong Port on the Tibet side, 261 of them foreign nationals.

How to report someone missing, including the cross-border and foreign-national routes.

How this briefing is put together

Casualty figures come from Nepal Police and NDRRMA situation reports. Damage, rescue and response figures come from the UN OCHA situation overview, 27 August and the latest NDRRMA update, which draw on NDRRMA, Nepal Police, the Department of Roads and the Nepal Electricity Authority. Different bulletins published at different hours give different counts; where two sources disagree, the lower confirmed figure is shown, with the source and the time named next to it. Every figure carries the time of the bulletin it came from, because a casualty count without a timestamp is not usable during an event.

If a number here is wrong, tell us and it gets corrected or removed. Every source this site uses is listed publicly.