This paper begins with a critical review of studies which have examined the effects of caring on health. Most are shown to suffer from defects in sampling and design, so that the evidence for detrimental effects is suggestive rather than conclusive. The substantive part of the paper then utilizes data on a cohort of 55-year-olds to compare the health of carers with the health of non-carers and to examine changes in caring and health over a 3-year period. The comparison yields no systematic evidence of the deleterious effects of caring on health; indeed, if there is a tendency in the accumulated data, it is in the opposite direction i.e. that carers report better health and functioning than non-carers. It is suggested that part of the explanation relates to selection and self-selection and the longitudinal data reveals high volatility in caring status, even over a short time period. The paper goes on to examine sub-groups of carers considered to be at greater risk. There is no evidence that their health is compromised but the authors acknowledge weaknesses in the data and argue for a specially designed study. The paper concludes with a discussion of the findings and their implications for research, policy and practice.