Blood haemoglobin concentrations and reticulocyte counts were determined in 430 school children in grades 1, 3 and 5 in a small township in rural Zambia at the end of the rainy season. Running capacity was estimated by the 12-min running test. Age, sex, height, weight, social and nutritional indicators were recorded. Mean haemoglobin was 125.3 g/1 +/- 12.3 SD. Low haemoglobin was associated with indicators of a traditional way of life. After a controlled trial of iron, folic acid, tetrachlorethylene and chloroquine lasting for 6 weeks, mean haemoglobin in all the children was 128.8 g/1 +/- 10.0 SD. The mean haemoglobin was significantly higher in the 218 children who received iron than in the 212 children who did not receive iron. There were no other significant differences among the treatment groups. In a subgroup of 377 children who completed two running tests, those with lower haemoglobin ran significantly shorter distances, even when possible confounding factors were partialled out. The effect was reduced by excluding 12 children with haemoglobin less than 100 g/1 or 48 children with reticulocyte counts of 15% or more, but remained statistically significant.