← history
Read stories →
AD 938

Battle of Bạch Đằng (938)

Ngô Quyền defeats the Southern Han fleet at Bạch Đằng

Wars & Conflicts Politics
Battle of Bạch Đằng (938) View full image ↗
Illustration generated by Ideogram 4 (fp8) (20 steps, 1216x832) on an NVIDIA RTX 5090 · 2026-08-27
How was this image made?

The creative brief sent to the image model. These are instructions, not a record of what actually happened.

Original prompt & settings (JSON) ↗

At high tide the iron-tipped stakes were invisible; hours later the falling river turned them into a barricade beneath the retreating fleet.

Late 938, in the Bạch Đằng estuary

Late in 938, a Southern Han fleet commanded by the ruler's son Liu Hongcao — Lưu Hoằng Tháo in Vietnamese — entered the Bạch Đằng estuary to intervene in Tĩnh Hải quân.

It was coming to rescue an ally who was already dead. Kiều Công Tiễn had killed Dương Đình Nghệ, seized power and requested Southern Han help; Ngô Quyền killed him before the fleet arrived.

The stakes and the tide

Ngô Quyền had iron-tipped wooden stakes driven into the riverbed. His smaller craft drew the Southern Han ships upriver while the high tide kept the barrier out of sight.

When the tide ebbed, his forces counterattacked and the retreating ships struck the stakes. The fleet was broken, Liu Hongcao was killed, and more than half the invading force was lost.

What the victory settled, and when

The Southern Han ruler Liu Gong abandoned his supporting advance and withdrew the survivors. The defeated power was Southern Han — a regional state based in present-day Guangdong.

The Red River region stayed outside Southern Han control. Ngô Quyền proclaimed himself king and established his court at Cổ Loa in 939, the following year.

No secure day survives: one scholarly account places the battle in autumn 938, while the Vietnam National Museum of History reconstructs the chronicle's "winter, tenth month" as approximately late November.

Sources

Researched 23 Aug 2026 5 sources 2 audit passes

How this was checked

Researched from the web into a fact sheet, rewritten from that sheet, then audited against it by a different model. How the pipeline works →

What the sources leave uncertain

  • No secure day is preserved. A scholarly account places the battle in autumn 938, while the Vietnam National Museum of History reconstructs the chronicle's 'winter, tenth month' as approximately late November 938.
  • Absolute troop strengths and casualty totals circulated online are not securely established; only the source's relative statement that more than half of the invaders were lost is retained.

Checked against