Hypovitaminosis A is considered to be the most common cause of blindness in the developing countries but it is not possible to estimate the prevalence of keratomalacia directly attributable to it. Subclinical hypovitaminosis A is not measurable at present in human subjects, but studies in animals indicate that the possibility of subclinical effects should not be ignored. The recommended procedure for identifying the "at risk" population involves a three-part survey to evaluate dietary intake, biochemical indices, and clinical signs. This article examines all three approaches in some detail, but in the present state of knowledge, none of them gives a satisfactory estimate of vitamin A status. For community assessment, the article discusses preliminary experience with a predictive model of the number of children in a population at risk of hypovitaminosis A that is based on associations noted repeatedly between protein-energy malnutrition and certain child-rearing practices, family economics, and morbidity. Criteria have been established for deciding on the need for a programme of prevention and the types of programme most appropriate in different situations are discussed. The methods of programme evaluation must take into account the stated objectives of the programme.