Past studies have shown that the more unaffordable housing is to people, the worse their health, particularly mental health. However, the commonly used housing affordability indicator, the 30% measure, has limitations. There is evidence that other indicators, including the 30/40 measure, might be more precise in characterizing housing unaffordability by taking into account absolute values of household incomes. In this paper, we use cross-sectional data from the UK Household Longitudinal Study, to evaluate relationships between two affordability measures (30%, 30/40) with 3 health measures: general, physical and mental health. We use logistic regression and effect modification to test whether relationships varied by age, ethnicity, housing tenure, urbanicity and sex. Out of 35,114 participants with complete data, housing was classified as unaffordable for 24.2% using the 30% measure and 10.2% for the 30/40 measure. In age-adjusted analyses, higher unaffordability was associated with worse health for all three health measures, with associations stronger for the 30/40 vs the 30% unaffordability measure. In models adjusted for age, sex and urbanicity, both ethnicity and tenure independently modified associations; with modification showing small differences by unaffordability and health measure. Further studies are needed to disentangle complex relationships between household income, housing costs, ethnicity and tenure.