During rapid-eye-movement sleep, when we dream, the brain is thought to be processing stored memory. The memory of a newborn infant is dominated by its fetal experience, and the infant is likely to dream about its life in the womb. Research with lucid (or conscious) dreaming has shown that dream images are supported by the corresponding body actions, using those muscles which remain active during rapid-eye-movement sleep. We suggest that sudden infant death syndrome or cot death may be a result of an infant dreaming about its life (or memory) as a fetus. In the course of that dream, since a fetus does not breathe (in the usual sense) the infant may cease to breathe and may die. This simple hypothesis is consistent with all of the known facts about sudden infant death syndrome (pathological and epidemiological), such as the age at death curve (the observed exponential decay and possibly the peak at 2-3 months), the higher risk with the prone sleeping position (but not excluding the supine position), and the observed climatic variation (seasonal and regional) in the incidence of sudden infant death syndrome. Many of these well-established facts have no other known explanation and other theories can generally only account for a few of the known facts about sudden infant death syndrome. Our hypothesis is also supported by recent findings that, as a group, sudden infant death syndrome infants have a higher proportion of rapid-eye-movement sleep, and also that they have an average higher heart rate (corresponding to possible fetal dreams) but only during rapid-eye-movement sleep.(ABSTRACT TRUNCATED AT 250 WORDS)