Mail surveys have been used to follow-up early participants in a Canadian breast cancer prevention trial. To minimize non-response bias, we undertook a randomized study of two postal strategies, of which one was our usual procedure and the other was a systematic application, known as the total design method (TDM) and described by Dillman. The response rates to the two methods were 62 and 88% respectively. The TDM is a practical, cost-efficient approach to reducing non-response bias in postal surveys and as such has an important role in epidemiological research which involves healthy participants.