They hope their discovery may prove useful in finding a treatment for severe malaria in children with other blood groups.
Their work is published in the Oct 30 issue of the Proceedings of the National Academy of Sciences.
The team, led by Alex Rowe of the University of Edinburgh in Britain, took blood samples from children with severe malaria, children with less serious malaria and healthy children in Bandiagara, Mali.
They found that children with group O blood were 66 percent less likely to develop severe malaria than children with A and B blood groups.
The blood of group O children contained far fewer clumps, or 'rosettes', of red blood cells than their non-O counterparts.
Rosettes appear when red blood cells infected with the malaria parasite - Plasmodium falciparum - stick to uninfected red blood cells, possibly to hide the parasite from the host's immune system.
Rosetting contributes to severe malaria by blocking the flow of blood through small blood vessels. "In the blood group O children there was very little rosetting at all," says Rowe.
"The study confirms for us the importance of rosetting in severe malaria. If we can modify rosetting so that we can make children who are A or B behave like those with group O - with very few rosettes - that could be a useful treatment for the most severely ill children," she adds.
The researchers aim to develop a drug to 'break up' rosettes in the blood of children with severe malaria. They already have a candidate compound, curdlan sulphate, but this can cause bleeding.
"We're trying to develop second-generation drugs based on curdlan sulphate that will still disrupt rosettes but not cause the bleeding," says Rowe.
"It looks like we'll be able to separate those two activities and have a drug that doesn't have those side-effects - though this is still very much ongoing work."