STUDY OBJECTIVE--The aim was to describe rates of loss and assessment of representativeness during 43 years of a national birth cohort study. DESIGN--The study population is a class stratified random sample of all single, legitimate births that occurred during one single week in 1946; it has been studied at regular intervals, so far to 1989. MAIN RESULTS--Losses through death and emigration were comparable to those in the national population of the same age. Response rates from the population resident in Britain have remained high, and the responding population is in most respects representative of the native population born in the early postwar years. Response rates within some serious physical illnesses did not differ from those of the healthy population. CONCLUSIONS--The continuing high response rate and representativeness of this national birth cohort is likely to be the result of home based data collections and of the regular contact to provide feedback of information and to check addresses of the study population.