Abstract Our reasons for this apparently single-minded viewpoint are simple. The technology of reliability assessment was worked out very early in the history of psychological assessment. We once traced its roots back to a textbook written in the 1930s, and the basic definitions were given without reference, indicating that, even seven decades ago, the classical definition was viewed as uncontroversial. There is complete consensus within the educational and psychological communities regarding the meaning of the term, and disagreement and debate are confined to a small cadre of biomedical researchers. Moreover, once one understands the basic concept it appears, in our view, to reflect perfectly the particular requirements of any measurement situation. Although there are some more recent theories in education and psychology, such as generalizability theory (see Chapter 9), these represent extensions of the basic approach. Other conflicting views which have arisen in the biomedical research community, such as the treatment of Bland and Altman (1986), offer, in our view, no advantage. This latter method, because of its recent popularity, will be discussed separately.