Objective - To assess the reliability of the VR-MICS/D (Verona-Medical Interview Classification System/Doctor) and to identify the verbal behaviour by general practitioners in interviews conducted with primary care attenders with medical complaints and emotional distress. Setting - Two general practices in South-Verona. Sample - 100 primary care patients attending for a new illness episode with a GHQ-12 score > 3. The five participating GPs contributed each with 20 audiotaped interviews of 10 patients judged by GP as emotionally distressed and of 10 judged without emotional distress. Main outcome measures — The VR-MICS/D classifies GPs' verbal behaviour during the medical interview into 16 categories in terms of form (question or statement) and content and allows to assess their interview skills. Procedure - Two raters classified 30 interviews (15 with patients judged by their GP as emotionally distressed and 15 with patients judged without emotional distress). Having established satisfactory reliability, the overall verbal performance, based on 100 interviews, was assessed and GPs' verbal behaviours with patients judged as emotionally distressed was compared with that adopted with patients judged without emotional distress. Results - The reliability was satisfactory (Kappa 0.93). Percentage agreements for categories varied between 78.2% and 96.4%. The most frequent verbal behaviours were closed ended questions and information giving (58% of a total of 5522 classified verbal units). Interviews with patients judged as emotional distressed contained a greater number of psychological and psychosocial contents, facilitating comments and clarifications. These differences, however, were small, despite their statistical significance. Conclusions - The VR-MICS/D is a reliable measure for describing GPs' verbal behaviour during the interview with emotional distressed patients. The interview style of the GPs in this study was similar to that reported in the literature for GPs without formal training in communication skills and was characterised by a prevalently doctor-centred approach. This approach, particularly with emotional distressed patients, has severe limitations and underlines the necessity of the introduction of communication skills training.