It is argued that a general practitioner's ability to make accurate ratings of psychological distress is partly determined by the rate at which patients emit cues that are indicative of such distress. This study addresses the behaviours of doctors which influence the rates at which patients emit such cues. Consultations were videotaped involving six General Practice Vocational Trainees, three of them poor, and three of them able identifiers of emotional illness. Consultations were selected so that each trainee was rated interviewing 4 patients with low GHQ scores, and 4 patients with high scores. Behaviours are described which lead to increased cue emission and which are also practised more frequently by able identifiers, while other behaviours reduce cue emission and are practised less frequently by them. Another set of behaviours is no more frequent among the able identifiers, but when practised by able identifiers is associated with increased cue emission by the patients, and when practised by poor identifiers with unaltered or decreased cue emission. Interviews that are 'patient-led' are associated with increased rates of cue emission, while those that are 'doctor-led' are associated with lower rates. The implications of these findings for training doctors working in general medical settings are discussed.