BACKGROUND: Early start of enteral feeding is an established treatment strategy in intubated patients in intensive care since it reduces invasive bacterial infections and length of hospital stay. There is equipoise whether early enteral feeding is also beneficial in non-intubated patients with cerebral malaria in resource poor settings. We hypothesized that the risk of aspiration pneumonia might outweigh the potential benefits of earlier recovery and prevention of hypoglycaemia. METHOD AND FINDINGS: A randomized trial of early (day of admission) versus late (after 60 hours in adults or 36 hours in children) start of enteral feeding was undertaken in patients with cerebral malaria in Chittagong, Bangladesh from May 2008 to August 2009. The primary outcome measures were incidence of aspiration pneumonia, hypoglycaemia and coma recovery time. The trial was terminated after inclusion of 56 patients because of a high incidence of aspiration pneumonia in the early feeding group (9/27 (33%)), compared to the late feeding group (0/29 (0%)), p = 0.001). One patient in the late feeding group, and none in the early group, had hypoglycaemia during admission. There was no significant difference in overall mortality (9/27 (33%) vs 6/29 (21%), p = 0.370), but mortality was 5/9 (56%) in patients with aspiration pneumonia. CONCLUSIONS: In conclusion, early start of enteral feeding is detrimental in non-intubated patients with cerebral malaria in many resource-poor settings. Evidence gathered in resource rich settings is not necessarily transferable to resource-poor settings. TRIAL REGISTRATION: Controlled-Trials.com ISRCTN57488577.