A 'patient-centred' approach is increasingly regarded as crucial for the delivery of high quality care by doctors. However, there is considerable ambiguity concerning the exact meaning of the term and the optimum method of measuring the process and outcomes of patient-centred care. This paper reviews the conceptual and empirical literature in order to develop a model of the various aspects of the doctor-patient relationship encompassed by the concept of 'patient-centredness' and to assess the advantages and disadvantages of alternative methods of measurement. Five conceptual dimensions are identified: biopsychosocial perspective; 'patient-as-person'; sharing power and responsibility; therapeutic alliance; and 'doctor-as-person'. Two main approaches to measurement are evaluated: self-report instruments and external observation methods. A number of recommendations concerning the measurement of patient-centredness are made.