This paper examines the problem of estimating indoor radon concentrations for radon-induced lung cancer risk assessment. Previous authors have identified various problems and possible corrections for the use of generally available radon screening measurements. Bias evident in data collected from volunteers is described; regional databases developed from voluntary data generally exhibit higher mean values than those obtained from random samples. Tools are developed to better characterize the differences in voluntary and random concentration distributions. Application is demonstrated using voluntary and random radon survey data for the state of North Dakota.