AbstractThis paper delineates the conditions whereby a sales manager would maximize long-run profits by using a typical two-step procedure to recruit salespersons. The procedure attempts to select the most able sales candidates from a flow of applicants, given the opportunity costs associated with the lack of validity and reliability of the procedure followed at each step. A computer algorithm (called SCRAP), based on the principles of Bayesian sequential sampling theory, makes the procedure operational. Required input data are generally available from existing personnel and accounting data, as well as from managerial judgements. The procedure is illustrated in a real situation and points the way towards more efficient recruiting practice.