Coffee is one of the first liquids given to infants in Guatemala. To evaluate whether this practice has an adverse effect on iron status, 160 children 12-24 mo of age who had received coffee for > or = 2 mo and had at least one indicator of iron deficiency were stratified by initial hemoglobin concentration (anemic, or nonanemic, ie, hemoglobin > or = 105 g/L) and randomly assigned to a control (continuation of coffee; coffee) or intervention (provided with a substitute consisting of sugar and coloring; substitute) group for 5 mo. Anemic children were provided with iron supplements for 2-3 mo. Hematologic and anthropometric measurements were made before and after the intervention and dietary and morbidity data were collected every 2 wk. A total of 139 children completed the study: 45 coffee, nonanemic; 56 substitute, nonanemic; 19 coffee, anemic; and 19 substitute, anemic. Compliance with the procedures was good: median coffee intake was 891 mL/wk in the coffee group compared with 18 mL/wk in the substitute group (P = 0.0001). There was no significant effect of discontinuing coffee consumption on changes in hemoglobin, hematocrit, ratio of zinc protoporphyrin to heme or plasma iron, zinc or copper in either nonanemic or anemic children, or plasma ferritin in children who did not take iron supplements. In children who took iron supplements, change in plasma ferritin was significantly greater in the substitute group than in the coffee group (106% compared with 1%, P < 0.05). This implies that coffee interferes with the utilization of supplemental iron. It is likely that the amount and strength of coffee consumed by Guatemalan toddlers are too low to significantly affect the other indexes of iron status.