STUDY OBJECTIVE--To provide a quantitative description of factors independently predictive of reported chickenpox infections in two national cohorts of British children. DESIGN--Longitudinal cohort study design employing logistic regression analysis of data obtained in the 1970 British Births Survey (later to become the Child Health and Education Study, CHES), and the 1958 British Perinatal Mortality Survey (later to become the National Child Development Survey, NCDS). SETTINGS--One-week birth cohorts covering the whole of the United Kingdom. PARTICIPANTS--Data were obtained from questionnaires administered to the carers of 10,196 children born in the UK between 5 and 11 April 1970 (CHES) and 10,927 children born in the UK between 3 and 9 March 1958 (NCDS). These numbers consist of the whole of the surviving cohorts excluding those for whom data were incomplete. MEASUREMENTS--Biological, social, and medical factors in the parents and children, as recorded by the child's principle carer or from clinical notes. MAIN RESULTS--Chickenpox by the age of 10 years was reported to be more common in the children of advantaged families (higher social class, higher parental education levels), with a higher prevalence in those parts of the United Kingdom normally associated with affluence, such as the South East and South West of England, and lower rates in Wales and Scotland. Chickenpox by 10 years was also associated with more crowding in the home. A similar but less marked pattern occurred for chickenpox by the age of 11 years in the 1958 NCDS cohort. This social distribution apparently reflected overall rather than age-specific susceptibility. CONCLUSIONS--The national and international pattern of chickenpox epidemiology indicate that both social and climatological factors may be important in defining groups at risk. Further research is indicated if a vaccination service is to be implemented in this country.