Results indicated that men who drop out of batterers' programs were more likely than men who remain in the programs to be unemployed; be unmarried, childless, or both; to have lower incomes; and to have less education. Dropouts were also more likely to have a criminal history, to report substance abuse or related problems, and to present with particular relationship concerns or orientations. In addition, the relationship between court referral and dropping out was inconsistent across studies and may also vary according to socioeconomic status. Psychopathology was consistently related to dropping out, but may be associated with other factors such as comorbidity or the referral source. Age, race, childhood exposure to violence, and battering history were all inconsistently associated with dropping out. Findings suggested the need for research using case reports, controlled experiments, quasi-experiments, surveys, and observational methods to gather crucial information on batterers' accountability. Prospective longitudinal studies would also be useful. Findings also suggested the need to consider screening for risk factors and to establish added strategies for program retention. Tables, notes, and 92 references (Author abstract modified)