The Uppsala Conflict Data Program records a best estimate of 25 deaths from organised violence within Angola in 2025, with a low estimate of 5 and a high estimate of 87.
The Uppsala Conflict Data Program (UCDP) at Uppsala University records a best estimate of 25 deaths from organised violence within the borders of Angola in 2025, in version 26.1 of its country-year dataset. Its low estimate is 5 and its high estimate is 87. The best estimate is the sum, over every event UCDP recorded, of the death toll its coders found most reliable in the source material; the low and high figures bound the reporting UCDP considered reliable at all.
UCDP attributes the best estimate to its three categories of organised violence as follows: 25 to state-based conflict — fighting over government or territory in which at least one party is the government of a state — with a low estimate of 5 and a high estimate of 78; none to non-state conflict — fighting between organised armed groups, neither of which is the government of a state; 0 to one-sided violence — the use of armed force by a government or an organised group against civilians — with a low estimate of 0 and a high estimate of 9. These are UCDP's categories and UCDP's attribution. This brief does not name any party to the violence.
In the same version of the dataset UCDP's best estimate for Angola in 2024 is also 25.
These figures are counts of deaths that UCDP's researchers identified in news reports and other published sources, and they are estimates, not a register of deaths. UCDP states that it is highly conservative in counting fatalities, and its country totals include only violence tied to a conflict or actor that has at some point reached its threshold of 25 deaths in a calendar year, so the figure is best read as a floor. It covers violence within the country's borders whoever the parties were, and says nothing about violence elsewhere in which the country's government took part.
This brief reports the figures UCDP published and nothing beyond them. It does not report who was killed, by whom, or why; it does not establish what any party to a conflict did or intended; and it does not attribute the figure, or any change in it, to any policy, event or decision.
Limits of this brief
UCDP's own annual estimate of deaths from organised violence within each country's borders, from the reviewed Country-Year Dataset on Organized Violence within Country Borders, with UCDP's low and high estimates beside every best estimate and the total attributed to UCDP's three categories: state-based conflict, non-state conflict and one-sided violence. The figures are annual and arrive about half a year after the year they cover, so this section is quiet for most of the year and then publishes a year's worth at once. These are estimates coded by researchers from news reports and other published sources, not a register of deaths. UCDP describes itself as highly conservative in counting fatalities and counts only violence tied to a conflict or actor that has at some point reached its threshold of 25 deaths in a calendar year, so every figure is best read as a floor, and country-years below that threshold are not published at all: a small total from this dataset does not mean what a reader would take it to mean. No brief names a party to the violence. UCDP publishes the names of the governments and groups involved, the deaths it attributes to each, and separate counts of civilians killed by governments and by armed groups; none of that is read here. UCDP's monthly preliminary event feed is not read either: its records are unreviewed by UCDP's own description, and an automatic account of who killed whom, where and when is reporting on disputed events that this service leaves to human editors. A comparison is only ever a country with its own preceding year, from the same dataset release, given as a difference in deaths and never as a percentage. UCDP revises earlier years in each release, so the previous year's figure on a brief is UCDP's current figure, not the one it published a year earlier.
If something here is wrong, write to
[email protected]
quoting the record identifier above. Corrections are recorded on this page,
not made silently.