Recent evidence suggests that plasma lutein is better correlated than either beta-carotene or lycopene with its respective carotenoid intake and therefore may be a better marker of vegetable intake than either beta-carotene or lycopene. In the study reported in this paper, measurements of plasma carotenes and retinol were made in infants from two villages near Peshawar in the North West Frontier Province, Pakistan, in July and November 1993. The approximate age at the start was 14 months, and 101 boys and ninety girls completed the study. Of the usual plasma carotenes, only lutein was measurable in all samples and was correlated with retinol in both boys (r 0.38, P < 0.0001; r 0.35, P < 0.001) and girls (r 0.21, P = 0.038; r 0.0307, P = 0.003) at the two time points respectively. In addition, the change in lutein was even more strongly correlated with the change in retinol in both boys (r 0.453, P < 0.0001) and girls (r 0.439, P < 0.0001). In August beta-carotene was measurable in approximately 8 % of samples and this increased to 31 % in November but there was no correlation between beta-carotene and retinol at any time. There were negligible amounts of lycopene and beta-cryptoxanthin in plasma at both time points. The mean concentration of plasma retinol in the infants was 0.66 mumol/l at baseline and 59% of the infants had retinol concentrations < 0.7 mumol/l. In addition, there were fifteen infants whose levels were below 0.35 mumol/l suggesting that vitamin A status in the population was marginal. Food intake of the infants was not monitored in the present study but breast feeding continues for up to 2 years in this part of Pakistan and most infants would be weaned onto selected foods eaten by the family. The close correlation between plasma lutein and retinol suggests that the increase in retinol over the summer season may be attributable to an increased availability of green vegetables to the families. The source of lutein to the infants is most likely to be the breast milk since such vegetables are unlikely to be given to infants except to suck as a weaning food. The results may indicate the potential usefulness of plasma lutein as a marker of changes in vegetable intake and changes in vitamin A status in Third World infants and children.