In order to identify homes with high radon levels as efficiently as po
ssible, it is necessary to have accurate numerical information on rado
n-prone areas. Maps based on geological indicators are generally quali
tative and sometimes contain significant errors. Maps based on radon m
easurement in homes are definitive if sufficient results are available
. Often there are small numbers of measurements, making it difficult t
o map the fraction of the housing stock which exceeds the reference le
vel. However, distributions of radon levels are usually log-normal, so
modelling of the distribution allows the fraction of the housing stoc
k exceeding any level to be calculated. A database of 62,555 radon res
ults in southwest England is used to show that the data within 5 km gr
id squares are consistent with log-normal distributions and to develop
techniques for estimating the fraction of homes exceeding a radon ref
erence level in each grid square, even where data are sparse.