This paper suggests that some of the difficulties of research utilisation in architecture stem from the complexity and diversity of architects needs. The architectural design process may be seen as proceeding through a series of sub-processes: briefing, design, implementation, evaluation. In each sub-process, designers and their collaborators generate a set of 'hypotheses' and then test them, drawing conclusions for further action. As this technical work is carried out, members of the design team must interact with each other and with 'outsiders', so naturally their behaviour affects the decisions which are made. In any of the sub-processes of design, architects and their collaborators may thus need information concerning possible 'hypotheses', concerning testing mechanisms or concerning the behavioural context within which they are working. In some instances they may need information of more than one kind. Clearly it would be of interest to the research community to establish a full empirical classification of the information requirements of each subprocess. This paper will therefore use case studies written over the last five years by the author and his colleagues to indicate the types of result which could be obtained. Interestingly these particular data suggest that social and psychological research is often not available when it is most needed.