Australian researchers examining the relationship between socioeconomic status and food-related behaviour have often selected their samples from the electoral roll and then collected their data using a mail-survey method. These studies have generally found statistically significant associations between socioeconomic status and behaviour, although these relationships are usually only weak-to-moderate in strength. Given the consistent and strong pattern of association between socioeconomic status and mortality, and diet and mortality, there is a possibility that these studies may have used a research design that underestimates the magnitude of the association. To assess this possibility, results obtained using an electoral-roll sample and mail-survey method were compared with findings obtained by administering the same questionnaire directly to a sample of indigent clients contacted through a welfare agency. The comparison suggests that studies that draw their samples from electoral rolls and then collect data using a mail-survey questionnaire may greatly understate the level of socioeconomic inequality in food-related behaviour in the wider community.