Most areas of health research will have accepted data and a dominating interpretation. If the interpretation is not correct, contradictions will accumulate, and it will eventually become clear that the current interpretation is untenable. In this situation, the best hypothesis is the one that accounts for all of the known data as well as the apparent contradictions. The area of vitamin A supplementation and childhood mortality in developing countries is afflicted with many contradictions and there is a need for a new hypothesis. We propose that the effect of vitamin A supplementation may depend on the amplification of non-specific effects of vaccines on childhood mortality.