There is an acknowledged problem with high non-completion rates of PhDs in the UK. Studies show that one of the major reasons for this is deficiencies in the supervision received. However, the unique and one-to-one nature of student/supervisor relationships makes it difficult to investigate the extent of these deficiencies. This article proposes a framework to evaluate the supervision requirements from the students' perspective. The supervision is divided into three elements: personal, indirect research-related and direct research-related help. The perceived short falls between expected and provided supervision for each element are then identified. A questionnaire survey is designed to act both as a test of the framework and to assess supervision requirements and their provision for PhD students in the construction management and engineering discipline in the UK. In addition, the student population is characterized and selected attributes; age, sex, industrial experience, full- or part-time status, nationality and funding sources, are compared with their supervision requirements. The results showed that the biggest discrepancy between expected and provided supervision was in direct research-related help. The framework is most valuable at individual level and for the framework to be effective communication is seen as being critical.