To determine the prevalence of vitamin A deficiency, infection and adequacy of vitamin A intakes among Zambian children, and the contribution of dietary vitamin A and infection to vitamin A status.

A cross-sectional survey of vitamin A intakes by the 24 h recall method, vitamin A status by plasma retinol and the modified relative dose-response test, and infection by acute-phase proteins.

Rural communities in Central and Eastern Provinces of Zambia.

Children 2–5 years of age.

The prevalence of vitamin A deficiency was 56 % by plasma retinol, 48 % with infection-adjusted plasma retinol and 22 % by the modified relative dose-response test. The majority of children (61 %) had a current infection. Vitamin A intakes were relatively high (331 to 585 μg retinol activity equivalents/d in the harvest/early post-harvest and late post-harvest seasons, respectively) and the prevalence of inadequate intakes was <1 % when compared with the Estimated Average Requirement (210 and 275 μg retinol activity equivalents/d for children aged 1–3 and 4–8 years, respectively). Elevated α-1-acid glycoprotein was negatively associated with plasma retinol (P < 0·0 0 1) and vitamin A intake was positively associated with plasma retinol (P < 0·05), but only when estimated assuming a 26:1 retinol equivalence for provitamin A from green and yellow vegetables.

Infection and vitamin A intakes were significant determinants of plasma retinol. We cannot conclude which indicator more accurately represents the true vitamin A status of the population. Reasons for the persistent high prevalence of vitamin A deficiency in the presence of adequate vitamin A intakes are unclear, but the high rates of infection may play a role.