Organised violence in 2026: what the data says

Counts of armed conflict come in two kinds that should not be mixed: verified data, which arrives weeks or months late, and news-based signals, which arrive within days but count something different. This article reads both, as of 17 October 2026.

The annual picture: 2025

The Uppsala Conflict Data Program (UCDP) publishes the previous year's figures around mid-year. Its 2026 release, dated 9 June 2026, covers 2025. According to the Uppsala University press release:

• 65 state-based conflicts, where states were involved on one or both sides, the highest number since data collection began in 1946.
• 8 interstate conflicts, the highest since 1946, up from 2 in 2023.
• 13 wars, meaning at least 1,000 battle-related deaths in a calendar year, the most since 1992.
• About 244,600 deaths from organised violence in total, the second-deadliest year since the 1994 Rwandan genocide.
• About 76,500 deaths in one-sided violence against civilians, more than four times the previous year and the highest since 1994.
• About 14,500 deaths in non-state conflicts, the lowest since 2013.
• The war between Russia and Ukraine was the deadliest conflict of the year, with at least 94,700 fatalities, about 62% of all battle-related deaths worldwide.

The PRIO paper Conflict trends: A global overview, 1946-2025, which uses UCDP data, reports 65 active conflicts in 35 countries and 75 non-state conflicts. It gives more than 255,000 fatalities, a higher total than the UCDP release's 244,600. The source pages do not explain the difference, so we report both and do not choose between them. PRIO also describes a sharp rise in violence against civilians, linked to Sudan.

The 2026 picture on Paravigil

Paravigil, ARMSMA's open-data platform, reads UCDP's monthly candidate data. Its data runs to 31 August 2026. On its overview page as of 17 October 2026, it shows 10,854 deaths in organised violence over 30 days, 12% more than the 9,680 in the period before, and UCDP's published figure for August 2026 is 11,118 deaths. Its analysis page shows 30,802 deaths in 61 countries over 90 days, with the worst-affected country accounting for 51% of them. That country is Ukraine; we discuss it at national level only.

Paravigil also shows model estimates for September 2026 (about 8,238 deaths, range 5,657 to 10,819) and October. These are estimates, not UCDP data. Paravigil states that its model, tested on 2023 to 2025, was off by 31% a month on average for the world total.

Verified data versus news signals

• Verified (UCDP). Analysts check each event against sources. Paravigil states that UCDP publishes monthly with a lag of 4 to 6 weeks, and the annual report comes months after the year ends. The data is slow but comparable over time.
• News-based (GDELT). Paravigil's overview counts 20,633 violence-related events in news reports over seven days. GDELT counts news reports, not incidents or casualties, so a rise can reflect more coverage. Paravigil advises comparing a country only with itself over time.

The two readings answer different questions: UCDP says how many people were killed, while news signals suggest where attention is turning now. Neither is a forecast, and Paravigil states that it is not an assessment by any government and not for emergency or operational decisions.