The analysis is based on a review of independent research in England and the United States. The cross-classification of adult offender typologies demonstrated the reliability of the groupings identified by cluster analyses of the Minnesota Multiphasic Personality Inventory protocols. Equivalent groupings emerged in independent analyses. The differences in the relative proportion of each type found in different samples, however, suggest that the groupings contained both a core group of similar individuals and also "less pure" or "outlying" examples of the type. The stability of the groupings in individual studies was not demonstrated; with one exception (homicide offenders), there was no indication of between-type differences on variables not included in the cluster analyses. In the absence of clearly defined types with established reliability and validity, there were limited options for the differential allocation of adult offenders to treatment or management conditions. It was equally evident in the review of young-offender typologies that the integrity of the types had not been established. 2 tables, 75 references.