This thesis sets out to describe and explain aspects of residential child care in New Zealand. It notes that the indigenous literature is sparse, that the subject has been neglected as a research area and that there is uncertainty about residential child care as a form of substitute provision. To clarify the subject area, a review of the international literature Focuses first on the way in which children are assigned to residential care. It then examines formative trends and influences, principles and methods and knowledge about the participants and suggests that the field lacks clear definition and identity. Prior to formulating a research strategy, previous studies are evaluated and five of these, dubbed the Milestone Studies, are examined in detail. Some proposals from General Systems Theory, in particular the study of complex adaptive systems, are reviewed and these elements built into a research model which draws together the constructs of perceptual process, role-interaction and system-matrix. This model develops six cumulative theoretical propositions based on those notions, defines its operational terms and formulates fourteen hypotheses for empirical testing. For the field work, a survey instrument, including a semantic differential device, was constructed and pilot tested. The population, consisting of staff members in certain roles plus all young persons aged thirteen years and over was reached through a two-stage postal survey. The First stage ascertained numbers and the second brought in 961 individual responses, estimated to be 80% of the total population as defined. The results confirm that although the two separate systems of residential child care, one run by the State and the other by diverse non-statutory agencies, have some characteristics in common, they are distinguished by the ethnic origins of their constituents, the length of time that members have been associated with child care and by different perceptions of roles and objects. Moreover, distinct intra-system differences between role-groups were observed. Differences are explained largely in terms of the emphasis of the State system upon the re-socialization of the adolescent offender compared with the non-State commitment to the younger, dependent child. This study concludes that it achieved the goals set, that it makes a small contribution to the research application of the semantic differential and that it provides a base for further study in the field of residential child care.