Objective: To assess vitamin A and E status and anaemia in non-pregnant Malawian adolescent girls. Design: A cross-sectional study in rural village communities in the Shire Valley, Southern Malawi. Subjects: Adolescent girls, n 118, aged between 10 and 19 y, 112 of whom were unmarried. Methods: Socio-demographic information was collected by questionnaire, heights and weights were measured. Vitamin A was assessed by the Modi®ed Relative Dose Response (MRDR) test, in addition to serum retinol values. Blood samples were collected 4±5 h after administration of 3,4-didehydroretinyl acetate. Retinol and atocopherol levels were measured by HPLC. Serum retinol results for non-pregnant girls were compared with values for 43 adolescent pregnant girls which were available from a previous study. Results: 26.6% of non-pregnant girls had serum retinol values`0.70 mmol/L; 40.2% had an MRDR ratio b 0.060. In 59.3%, serum tocopherol levels were`11.5 mmol/L and the tocopherol/cholesterol ratio was`2.2 in 23.9%. 11.3% had a haemoglobin ! 12 g/dl. Vitamin A levels were signi®cantly related to age, and younger girls were more likely to be de®cient. Signi®cant correlations were found between serum retinol, MRDR values and serum tocopherol. Girls with a low body mass index for age had tocopherol cholesterol ratios`2.2. Low serum retinol values occurred signi®cantly more often in stunted girls (P 0.01). Serum retinol values of adolescent pregnant girls were signi®cantly lower than those of non-pregnant adolescents (P 0.002). Conclusions: Vitamin A and E de®ciency and anaemia were common in adolescent non-pregnant girls, and thought to partly result from increased growth requirements. Girls who become pregnant at an early age are at risk of depletion of their nutritional reserves. Measures to reduce nutritional de®ciencies before the ®rst pregnancy are needed.