Self-report postal questionnaires have been advocated as an efficient means of collecting local data on adult oral health needs. The aim of this study was to compare the response from deprived and affluent communities and examine a method for the detection and compensation of non-response bias. An oral health status questionnaire was administered by post to random samples of older residents from affluent and deprived electoral wards. The survey was conducted in three distinct stages to increase the response rate and to quantify the effects of non-response. A response of 59.6 per cent was achieved from the deprived ward and 77.7 per cent from the affluent ward, this difference was statistically significant. The response rate compared favourably with contemporary national and international studies of oral health using postal survey methods. The pattern of response over the three stages was used to detect the presence and direction of non-response bias. This analysis showed evidence of non-response bias for only one variable of interest, for which an estimated prevalence value was calculated.