EMPHASIS IS PLACED ON PREDICTIONS FORMULATED WITH THE PURPOSE OF MAKING DECISIONS AND, IN PARTICULAR, ON PREDICTIONS FOR USE IN PENAL DECISIONS AFFECTING INDIVIDUALS. SOCIAL PREDICTIONS ARRIVED AT IN A RATIONAL MANNER I.E., PREDICTIONS, THE THEORETICAL BASIS AND EMPIRICAL OBSERVATIONS OF WHICH ARE EXPLICITLY STATED SO THAT THE PREDICTIONS MAY BE REPLICATED, AND THEIR METHODOLOGY ARE THE SUBJECT OF THE STUDY. CONTEMPORARY PRACTICE IN CRIMINOLOGICAL PREDICTION IS DISCUSSED, PROBLEMS OF PREDICTION ARE IDENTIFIED, AND REQUIREMENTS WHICH PREDICTION INSTRUMENTS MUST FULFILL ARE FORMULATED. STATISTICAL TECHNIQUES FOR MAKING PREDICTION INSTRUMENTS ARE PRESENTED. AN EXAMINATION OF THE ROLE OF SCIENTIFIC KNOWLEDGE PREDICTION AND EXPLANATION GIVES RISE TO A DESCRIPTIVE MODEL OF EXPLANATION AND PREDICTION. THE DESCRIPTIVE MODEL IS EXTENDED TO CASES OF INDETERMINISTIC KNOWLEDGE WHERE THE SUBJECT OF PROBABILITY COMES TO THE FORE. THE DIFFERENT TYPES OF PROBABILITY ARE IDENTIFIED AND DISCUSSED, AND TWO CONCEPTS ARE SELECTED FOR CLOSER SCRUTINY: PROBABILITY AS A SUPPORT-RELATION, AND PROBABILITY AS AN ABSTRACT AXIOMATIC STRUCTURE WHICH MAY BE USED TO REPRESENT RELATIONSHIPS IN EMPIRICAL SITUATIONS. THE RELATION BETWEEN INDUCTIVE AND MATHEMATICAL CONCEPTS OF PROBABILITY IS STUDIED. THE IMPORTANCE OF PROBABILITY AS AN APPROACH TO CAUSALITY IS STRESSED, AND AN EVALUATION OF THIS INTERPRETATION FOR PREDICTION OF DELINQUENCY IS PRESENTED. NOTES FOLLOW EACH CHAPTER, AND A BIBLIOGRAPHY IS PROVIDED. (DAS)