The purpose of this study was to predict maintenance of foster home placement. Fifty-one foster children comprised the sample; 47 were involved in the Casey Family Program in Montana, and four were past participants. Case files provided the first and major data base. Caseworker ratings, a second data source, were obtained on children's personal characteristics, foster family characteristics, and factors related to foster placement breakdown. The third data source involved six categories of archival data regarding foster children (for instance, placement history, risk factors, and relationship factors). Factor analysis of the data produced 324 items pertinent to predicting foster home maintenance. These items were grouped into seven domains: personal and demographic characteristics, natural family history, relationships and developmental factors, placement history, present foster family characteristics, and factors involved in placement breakdown or change. Outcome variables included occurrence of placement breakdown, as well as caseworker ratings of three factors: the child's level of functioning, the foster family's level of functioning, and the likelihood of choosing specific foster families for future placements. Results indicated foster mothers with strong emotional coherence generally correlated with high ratings of foster children's overall functioning and with stable placements. Placement breakdown was highly predictable when foster child and foster family characteristics were considered. (Author/DB)