Using the World Health Organization criterion, the prevalence of sub-clinical vitamin A deficiency can be assessed using plasma retinol concentrations <0·7 μmol/l. However, plasma retinol can be depressed by infection; thus, the use of this criterion alone may overestimate deficiency. In the present study, we investigated the usefulness of the acute-phase proteins (APP) α1-antichymotrypsin (ACT) and α1-acid glycoprotein (AGP), plasma carotenoids and anthropometric and gestational indices to interpret plasma retinol in the blood of 192 apparently healthy Nigerian neonates collected randomly during days 1–20 postpartum. The mean weight (2·64 kg) and length (0·458 m) of the neonates and plasma concentrations (geometric mean, μmol/l) of retinol (0·54), α-carotene (0·072), ß-carotene (0·076) and lutein (0·080) were low. The prevalence of vitamin A deficiency was 72 %, indicating a severe public health problem. Babies who were of low birth weight (P<0·003) or premature and low birth weight (P<0·023) had significantly lower retinol concentrations than full-term normal weight babies. Thirty-two neonates had abnormal ACT and forty-four abnormal AGP concentrations. Positive correlations between retinol and ACT (r 0·186, P=0·05) and AGP (r 0·31, P=0·0001) during days 1–5 may be due to the increasing plasma retinol from maternal milk and a coincidental increasing capacity to synthesise APP. Subsequently, negative correlations between retinol and ACT (r −0·28, P=0·02) and AGP (r −0·29, P=0·018) from day 6 onwards reflected the continuing increase in plasma retinol, but no further increase in the APP. Overall, weight, ACT, lutein and age explained 30 % of the variance in retinol, but lutein was the most significant (r2 0·18, P<0·0001). Hence, the distribution of plasma retinol concentrations in this group of neonates was more strongly linked with nutrition (via the surrogate marker lutein) than infection.