The participants came from two cities and were recruited by newspaper advertisement and bulletin board posting in two shelters for battered women. The participants completed instruments and interviews focusing on education, relationship control, relationship duration, childhood abused, depression, self-esteem, physical symptoms, and self care. The original sample included 97 women who were battered and 96 who were not battered, as determined by an adaptation of the Conflict Tactics Scale. Of the 114 who returned approximately 2.5 years later, only 25 percent of the 51 originally battered women were still battered. The data were analyzed using discriminant function analysis. Results revealed no significant differences between those who were battered and those who were not at the second interview. Results do not support a learned helplessness model for most women experiencing abuse. Findings also suggest that abused women seen in the health care system, social services system, or both should be supported in a process of healthy decisionmaking about their relationship status. Tables and 39 references (Author abstract modified)