This paper presents a critique of current methods employed to assess the motivation of rehabilitation patients, and points out the problem faced by the psychologist who is expected to motivate the problem patient. An alternative technique based upon Premack’s research is suggested, in which participation in therapy classes which are attended with a high probability is used as a reinforcer by making it contingent upon attending classes of lower probability. Two cases are described in which such a model was successfully employed. Some possible problems involved in implementing this approach are discussed.