SynopsisInterpersonal relationships pose problems additional to those arising from social behaviour. If there is to be a science of interpersonal relationships, it must rest on a firm descriptive base. It is suggested that the following categories of dimensions are likely to prove useful: the content of the interactions within the relationship; their diversity; their qualities; their relative frequency and patterning; the nature of their reciprocity v. complementarity; and the intimacy, interpersonal congruency and commitment between the participants. A science of relationships will also require principles concerned with their dynamics. Three likely categories are those concerned with social constraints, with learning, and with positive and negative feedback. Principles concerned with learning will require an adequate categorization of the resources used in interpersonal interactions. The specification of these principles depends on the adequacy of the descriptive base.