The objective of the present study was to determine marginal vitamin A deficiency (VAD) by testing the hydrolysis of retinoyl glucuronide (RAG) to retinoic acid (RA) in children. Previous studies in rats showed that hydrolysis occurred when rats were vitamin A deficient. Children (n 61) aged 3–18 years, were divided into two groups, I and II. Blood was collected from the children in Group I (n 19) who were not dosed with RAG. Children in Group II (n 42) were administered all-trans retinoyl glucuronide (RAG) orally, and blood was collected 4 h after the dose. All serum samples were analysed for retinoids and carotenoids. RA was detected in serum only when serum retinol was < 0·85 μmol/l. Thus, hydrolysis of RAG to RA occurred in children with VAD or marginal VAD. Serum retinol was < 0·35 μmol/l in twenty-one children, 0·35–0·7 μmol/l in twenty-three children, 0·7–0·9 μmol/l in eleven children and >1 μmol/l in six children. Mean serum retinol in sixty-one children was 0·522 (sd 0·315) μmol/l. Mean β-carotene (0·016 (sd 0·015) μmol/l) was far below normal compared to the level of lutein (0·176 (sd 0·10) μmol/l) in sixty-one children. A low β-carotene level might be due to a low intake of carotene but high demand for vitamin A. The RAG hydrolysis test may prove to be a useful approach for the determination of marginal VAD with no clinical or subclinical signs of VAD. High prevalence of VAD amongst certain communities in Assam cannot be ruled out.