OBJECTIVE: To explore GPs' experience of carrying out 'talking therapy'. METHODS: Qualitative study using semi-structured interviews with 11 Danish GPs sampled purposively. The material was analysed by Interpretative Phenomenological Analysis. RESULTS: The participants expressed difficulty in explaining how they carried out talking therapy. However, from their description of individual therapies their perception of important aspects of methodology could be obtained: (1) their own open receptiveness, e.g. attentive listening, not limited by time; (2) relational factors including trust and empathy developed over time, or more active therapeutic use of the relationship; (3) knowledge of the patient's life story, told or written, used to form a model of the patient's problems, thoughts and feelings. The sessions were not offered if the GPs lacked time. CONCLUSION: Participants were mostly self-taught and did not use specific methods systematically despite having learnt them. GPs knew the patients beforehand; talking therapy developed from other treatment, and methodology had to fit into this. Specific methods are possibly not relevant in general practice. PRACTICE IMPLICATIONS: Formulation of a theory of talking therapy based on the views and experience of GPs and including non-specific factors could professionalize the field.