Objective: We investigated whether there was a growth or morbidity response to zinc supplementation. Design: The study was randomized, placebo-controlled, and double-blind. Setting: Children were recruited at clinics in Kingston, Jamaica, and supplemented at home. Subjects: Children selected were singletons aged 6±24 months, and stunted (`À2.0 s.d. length for age, NCHS references). They were strati®ed by sex and age and randomly assigned to receive zinc supplement (n 31) or placebo (n 30). Four children were excluded because of hospitalization; all others had all measurements. Adequately nourished children (n 24) were recruited from a well-baby clinic. Interventions: The supplement provided 5 mg elemental zinc in a syrup daily for 12 weeks; the placebo comprised the syrup only. Main outcome measures: Caretakers were interviewed to obtain social background data, number of clinic visits and hospitalizations. Anthropometric measurements were done on enrolment, and after 6 weeks, 12 weeks and 12 months. Children's health was determined by weekly questionnaire to caretakers of the undernourished groups during the supplementation period. Results: The supplemented and placebo groups were similar on enrolment. The adequately nourished children were from signi®cantly better socio-economic circumstances. Mean initial hair zinc content was 5.5 AE 4.8 mmolag (supplemented group) and 6.7 AE 12.1 mmolag (placebo)(n.s.). Regression analyses showed that there were no signi®cant effects of supplementation on length, height or head circumference, nor on the incidence of any morbidity symptom. Mean duration of the episodes was signi®cantly shorter for skin rashes in the supplemented group compared with the control group (ANCOVA, P 0.02), and longer for vomiting (P 0.02). The incidence of hospitalization was signi®cantly greater in the control group (Fisher's exact test, P 0.02). Conclusions: Zinc supplementation reduced the hospitalizations which probably re¯ect severity of morbidity, but did not improve growth.