This is a prospective study involving two centres. Children younger than 16 years old who had severe malaria underwent funduscopy. In one centre, direct ophthalmoscopy was performed by both an ophthalmologist and a non-ophthalmologist physician. At the other, two ophthalmologists performed a funduscopy on each patient using different methods (one using direct and the other indirect ophthalmoscopy). The objective was to determine the frequency of retinopathy and evaluate the effectiveness of direct ophthalmoscopy in detecting retinopathy in constantly dilated eyes. Of 62 children seen at both centres, 37% were found to have malarial retinopathy. Macula whitening was the most common finding. There was substantial agreement between the ophthalmologists using either the direct or indirect ophthalmoscopy techniques (Kappa value 0.604). However, at the centre using only direct ophthalmoscopy, there was poor agreement between the ophthalmologist and the non-ophthalmologist (kappa value -0.244). Direct ophthalmoscopy seemed to be effective in detecting malarial retinopathy, but only when performed by experienced personnel.