Yemen's Red Sea Offensive Displaces More Than 100,000

At least 100,000 people have fled their homes after renewed fighting between government forces and the Houthis, according to the UN refugee agency Al Jazeera.
The displacement follows a Houthi assault in the week before September 16, 2026 that captured the whole Red Sea coast, including the port city of Mocha. Government forces carried out air attacks on Houthi positions in an attempt to retake Mocha days after the city was captured.
Taiz, a governorate like a province in southwestern Yemen, received most of the displaced people. Ali al-Shumairi, director-general of the Executive Unit for the Management of IDP Camps in Taiz, said 65,845 people were displaced in Taiz governorate between August 10 and September 15, 2026. IDP stands for internally displaced person, a civilian forced to flee home but still inside the country. About 95 percent of those displaced in Taiz came from five districts: Jabal Habashi, al-Maafer, Maqbana, al-Wazi'iyah and Mawza.
Al Jazeera published first-hand field reporting from Taiz on September 16, 2026, including interviews with displaced people and the Taiz IDP camps chief and photos of newly displaced persons at al-Munij camp credited to Al Jazeera. Among those interviewed in Taiz was 70-year-old Hajja Fatima Mohammed Saleh, displaced from Maqbanah district during fighting between government forces and Houthis. On September 10, 2026, displaced civilians fleeing clashes in Al-Wazi'iyah district were sheltering in temporary shelters in the Bani Omar area of southern Taiz.
Later counts reported higher numbers. On September 4, 2026, Al Jazeera reported escalating Houthi attacks on Taiz, with intensifying government-Houthi clashes displacing thousands and threatening roadways to the Red Sea coast Al Jazeera. On September 7, the International Organization for Migration reported 18,500 people displaced in 72 hours amid renewed hostilities along Yemen's western coast and in parts of Taiz and Al-Hodeidah IOM. IOM is a UN-linked agency that assists migrants and displaced people. Civilians were reportedly among those killed and injured as clashes intensified in those areas in early September.
Tracking through early and mid-September recorded larger movements over time. Between August 30 and September 5, 2026, IOM Yemen DTM tracked 2,250 households (13,500 individuals) displaced at least once, within a total of 5,063 households (30,378 individuals) tracked from January 1 to September 5. DTM, the Displacement Tracking Matrix, is IOM's system for counting displacement, and a household means people living together as a family unit. At least 5,688 households were reported displaced across Taiz and southern Al Hodeidah between September 3 and 7. Between September 6 and 12, DTM tracked 9,892 households (59,352 individuals) displaced at least once.
Subsequent agency totals moved higher within days. The International Organization for Migration reported that displacement topped 76,000 as fighting intensified ReliefWeb. IOM said Taiz was the hardest hit with 8,300 households, or nearly 50,000 people, displaced from districts including Jabal Habashy, Al Ma'afer and Maqbanah. A UN report on September 11 cited IOM figures of at least 46,000 displaced since renewed fighting the previous week between Houthi rebels and pro-government forces in southwestern Yemen. Countrywide, some 11,442 households were newly displaced since hostilities escalated on September 3, including 10,310 households across 15 districts in the Taiz hub. Reuters reported 823 families displaced in Taiz in a single day in September 2026, and on September 14 published a report filed from Taiz on displaced Yemenis amid the Houthi advance Reuters, describing the Houthis as Iran-aligned.
The broader context here is concentration. Nearly all recorded movement in Taiz originates in a small cluster of districts, which focuses shelter demand on a limited set of camps and host sites such as al-Munij and Bani Omar. That pattern increases the risk of secondary displacement if front lines shift again around Mocha and the coastal roadways, where air attacks have so far not reversed territorial loss.
In my view, practitioners should read the sequence less as competing totals than as a single escalation curve captured by different instruments, like separate cameras photographing the same crowd from different angles. Executive Unit registrations, RDT event snapshots and IOM consolidated estimates each measure slightly different populations over slightly different windows. Together they point to sustained outflow from late August into mid-September, centered on Taiz and southern Hodeidah, with the contest for the Red Sea coast as the immediate driver of movement.


