Limited adherence to iron supplementation is thought to be a major reason for the low effectiveness of anemia-prevention programs. In rural Tanzania, women at 21-26 wk of gestation were randomly given either 120 mg of a conventional (Con) iron supplement or 50 mg of a gastric-delivery-system (GDS) iron supplement for 12 wk. Adherence was assessed by using a pill bottle equipped with an electronic counting device. Adherence in the GDS group was 61% compared with 42% for the Con group. In both groups, women experiencing side effects had about one-third lower adherence. Fewer side effects were observed in the GDS group. In a subgroup of women with a low initial hemoglobin concentration (< or = 120 g/L), the response to the iron supplements suggested that both of the applied doses were unnecessarily high for adequate hematologic response in a population with a marginal hemoglobin concentration. The GDS group appeared to require a dose one-fourth as high as that of the Con group for an equal effect on improving hemoglobin to normal concentrations.