The advent of victimization surveys has permitted much greater specificity in measuring differences in criminal victimization probabilities among various subgroups in the U.S. population. However, the availability of more data raises the question of whether differences in the probability of victimization truly depend on differences in the amount of exposure to crime by various segments of the population. Nonetheless, both commonsense and existing data clearly show that probabilistic exposure to crime is not a random process (i.e., not all persons, objects, times, or places are equally probable targets for a criminal offense). Data on persons who have experienced multiple victimization also support the notion of the nonrandom nature of probabilistic exposure to crime. An inferential link between the characteristics of victim-prone people and criminal victimization through the concept of exposure can be made by a lifestyle/exposure model of personal victimization. This basic model postulates that role expectations and social structure impose constraints to which persons commonly adapt in society. These role expectations and structural constraints for any individual depend upon that individual's constellation of demographic characteristics. However, major deficiencies in existing data that impede greater predictability of probabilistic exposure include the lack of precision in measurement of situational characteristics of criminal incidents, which may enhance risk, and the absence of refined and direct measures of lifestyles (e.g., changes in lifestyle with respect to age and marital status). Two important theoretical directions in victimization measurement are typologies of victim proneness and the role of offender motivation. Twenty-five references and footnotes are provided.