It has been suggested that socio-economic gradients in health reduce or disappear during youth, to be re-created during early adulthood through a process of health-related social mobility. The present analysis tests this hypothesis in relation to 'medical mortality', using a data set which is free of numerator-denominator bias. The sample consists of the appropriate age groups in the OPCS Longitudinal Study; 62,647 males and 59,644 females aged 0-14 at the 1971 census. 'Medical mortality' during 1971-1985, calculated as standardised mortality ratios, is analysed by parental social class, housing tenure and car access in 1971. 'Medical mortality' during 1981-1985 is analysed by own social class in 1981. The results suggest that 'medical mortality' is inversely related to social advantage at ages of death 0-9 years, that this gradient flattens or disappears at ages 10-14 and that it re-emerges at ages 15-29. Within the present analysis this apparent re-emergence could not have been due to health-related social mobility. It is concluded that the apparent absence of socio-economic gradients in 'medical mortality' during youth may be an artefact of the high levels of health enjoyed by this age group and its consequent low levels of non-accidental death.