The prevalence of vitamin A (VA) deficiency, which affects about one-third of children in developing countries, is falling only slowly. This is despite extensive distribution and administration of periodic (4- to 6-monthly) high-dose VA capsules over the past 20 years, now covering a reported 80% of children in developing countries. This massive programme was motivated largely by an expectation of reducing child mortality, stemming from findings in the 1980s and early 90s. Efficacy trials since 1994 have in most cases not confirmed a mortality impact of VA capsules. Only one large scale programme evaluation has ever been published, which showed no impact on 1-6-year-old mortality (the DEVTA trial, ending in 2003, in Uttar Pradesh, India). Periodic high-dose VA capsules may have less relevance now with changing disease patterns (notably, reductions in measles and diarrhoea). High-dose VA 6-monthly does not reduce prevalence of the deficiency itself, estimated by low serum retinol. It is proposed that: (i) there is no longer any evidence that intermittent high-dose VA programmes are having any substantial mortality effect, perhaps due to changing disease patterns; (ii) frequent intakes of vitamin A in physiological doses -e.g. through food-based approaches, including fortification, and through regular low-dose supplementation-are highly effective in increasing serum retinol (SR) and reducing vitamin A deficiency; (iii) therefore a policy shift is needed, based on consideration of current evidence. A prudent phase-over is needed towards increasing frequent regular intakes of VA at physiological levels, daily or weekly, replacing the high-dose periodic capsule distribution programmes. Moving resources in this direction must happen sooner or later: it should be sooner.