The moral questions that arise in this area are derived from the concepts of freedom and democracy and the balance between individual and State rights and balance between individual and State rights and privileges. These abstract ideas take shape in concerns as to the appropriate uses of information about persons who have been found guilty of crimes, as well as those who are merely accused of offenses. If discrimination between certain persons in terms of particular attributes (e.g., ethnic origin) is forbidden, then it may be argued that the use of such information is morally unacceptable, even if it is not illegal. If an unacceptable type of information is included in equations, its use may be hidden among other data. This may present a problem, but the final use of the conglomerate data should, perhaps, provide the criterion. For this reason, it is argued that the most efficient prediction equations should be sought, but only to test for any loss of efficiency due to the exclusion of certain data, to estimate the probable cost of using the simpler methods. In addition to issues that arise in regard to the kinds of information and their uses, there are concerns related to the offender's legal status. Persons who have been found guilty are usually assumed to have lost some of their constitutional rights. It is argued that one of the rights derogated is the right to withhold personal information. Thus, the moral issues surrounding the use of predictive techniques for persons not found guilty (and therefore presumed innocent) are different from those raised by the prediction of recidivism, and prediction of recidivism differs morally from the prediction of probable delinquency. The concepts of due process and just deserts are necessary and provide useful constraints on the uses of predictive instruments. Five references are provided. (Author abstract modified)