Radon, a well-established risk factor for human lung cancer, is present at low concentrations in most homes. Consequently, many countries have established national guidelines for residential radon concentrations. In this article, we evaluate two models for describing seasonal variation in residential radon concentrations based on the data from a large case-control study conducted in Winnipeg, Canada. In this study, radon levels in homes were monitored during two successive 6-month periods, with integrated annual radon concentrations obtained using CR-39 alpha-track detectors. Significant differences were noted among measurements taken during different seasons of the year. Using the model introduced by Pinel et al. (1995) to describe temporal variation in residential radon levels in southwest England using seasonal adjustment factors, reasonable predictions of annual average radon concentrations were obtained from the 6-month integrated radon measurements. However, a simple multiplicative model was found to provide better predictions than the seasonal adjustment model. Although model coefficients vary somewhat from one geographic location to another, the concordance with respect discriminating between results above and below 150 Bq/m(3) in Winnipeg was in the range 85-90% using seasonal adjustment models with coefficients derived from data in either Winnipeg or southwest England.