Thirty-two ewes were fed retinyl propionate equivalent to 0, 120, 1200 or 12,000 micrograms/kg body weight from 75 d of gestation. At 11 +/- 2 d before parturition 2 to 4 ewes per group were given intravenously 3H-retinol. Blood was sampled sequentially. Following parturition, ewes were milked twice daily for 7 days. Concentration and specific activity of vitamin A were determined in milk and plasma samples. Four or five exponential terms were necessary and sufficient to fit the labelled vitamin A disappearance from plasma with time. Calculations from these equations showed that retinol transport and clearance were similar to values observed earlier in gestation. Milk volume was unaffected by treatments. Milk concentration of vitamin A was correlated with intake. Concentration was maximal in initial milk secretions and decreased with time. Two exponentials were necessary and sufficient to fit the appearance of tracer vitamin A in milk. These curves were unaffected by treatment. Milk specific activities were approximately equal to corresponding plasma specific activities in ewes fed control intakes of vitamin A, indicating that the origin of milk vitamin A was plasma retinol-retinol-binding protein. Appearance of tracer in milk correlated with plasma retinol clearance; this suggests that secretion of vitamin A from plasma into milk is not regulated. These findings collectively show that maternal vitamin A status directly determines the supply of vitamin A to the neonate.