The war was declared over in December 2002, but fighting continues
The number of people killed in a decade of war in the Democratic Republic of Congo may be half of the accepted toll of 5.4 million, a study has suggested.
The Canada-based Human Security Report Project says the figure makes too many assumptions about how many deaths were caused by the war.
The researchers say many of the deaths between 1998 and 2008 would have occurred without conflict.
The figure was used to justify ramping up the UN presence in the country.
The BBC's East Africa correspondent Peter Greste says the initial figures shocked the world into action.
A formal peace accord ended a war that had involved Angola, Namibia, Zimbabwe, Uganda and Rwanda in December 2002, but unrest has continued in the volatile east of the country.
The 5.4 million toll was calculated by the aid agency International Rescue Committee (IRC).
Rick Brennan, one of the authors of the original IRC study, acknowledged some statistical problems with the research.
But he said their methods and figures had been widely reviewed and generally accepted as a fair estimate of the number killed either as a direct result of fighting or because of disease and malnutrition.
The IRC assumed that the number of people dying each year in DR Congo in peacetime would be similar to rates elsewhere in other parts of sub-Saharan Africa.
It then analysed how many people had died from 1998 to 2008 and attributed the difference in the two figures to the conflict.
But the new study argues that the rate is unrealistic.
The researchers say large numbers of people would have died without the conflict - simply because basic living conditions in DR Congo were so tough.
When the researchers used their higher mortality rate to recalculate the figures, they found the number dropped below 3 million.