The need for regular up-dating of the selection probabilites in continuous surveys is emphasized in this paper. A simple strategy (selection method for the initial sample with the revision procedure) is presented and its application to the Canadian Labour Force Survey is discussed.