Data collected as a part of a survey on radon concentrations from about 40 000 dwellings in England for six contrasting geological units were analysed to evaluate the impact of house-specific factors (building characteristics and construction dates) and of proximity to geological boundaries. After adjusting for temperature and outdoor radon, geological unit, house type, double glazing and date of building were found to have a statistically significant influence on indoor radon concentrations and explained about 29 % of the total variation between dwellings in logarithmically transformed radon values. In addition, there were statistically significant differences in radon concentrations according to proximity to geological boundaries categories for most of the geological units, but no consistent pattern could be detected.