Food-based approaches for controlling vitamin A deficiency and its consequences, such as increased mortality, more severe morbidity, and anemia, have become increasingly important, thus prompting a reassessment of the relation between vitamin A intake and status. A nutrition surveillance system in Central Java, Indonesia, assessed the vitamin A intake and serum retinol concentration of women with a child < or =24 mo old with a semiquantitative 24-h recall method that categorized vitamin A-containing foods into 3 categories of plant foods and into 2 categories of animal foods and identified portions as small, medium, or large. Median vitamin A intake was 335 retinol equivalents (RE)/d (n = 600) and vitamin A intake from plant foods was 8 times higher than from animal foods. Serum retinol concentration was related to vitamin A intake in a dose-response manner. The multiple logistic regression model for predicting the chance for a serum retinol concentration greater than the observed median (> or = 1.37 micromol/L) included physiologic factors, vitamin A intake from plant [odds ratio (95% CI) per quartile: 1st, 1.00: 2nd, 1.23 (0.75, 2.02); 3rd, 1.60 (0.97, 2.63); and 4th, 2.06 (1.25, 3.40)] and animal [1st and 2nd, 1.00; 3rd, 1.31 (0.86, 2.02); and 4th, 2.18 (1.40. 3.42)] foods, home gardening [(no, 1.00; yes, 1.71 (1.12, 2.60)], and woman's education level [< or =primary school, 1.00; > or =secondary school, 1.51 (1.02, 2.22)]. Despite the fact that plant foods contributed 8 times as much vitamin A as did animal foods, serum retinol concentrations did not reflect this large difference. Home gardening and woman's education level seemed to reflect longer-term consumption of vitamin A-rich plant and animal foods, respectively.