The two groups of mothers were matched for age of baby, race, and socioeconomic status. Subjects were classified as abused or not abused based on their responses to the Physical Punishment Scale of Assessing Environments III and on information provided during structured interviews. Mothers also completed the Beck Depression Inventory. Free response methods were chosen for the research because they were well-suited to abused mothers who tended to be sensitive about their histories and parenting abilities. Results showed that clusters of negative attributes pervaded memories and perceptions abused mothers had of others, particularly parents. Moreover, the degree of negative elaboration, i.e., the number of negative clusters attributed to others, discriminated abused and control groups almost perfectly. It was also found for both groups that the more elaborated the positive view of self and others, the more secure the attachment of infant to mother. In addition, patterns of identification and description were dramatically different between the two groups. Unlike control mothers, abused mothers tended to disidentify with their own mothers and to be inconsistent in their characterization of them. 42 references