BACKGROUND: In 2024, one-third of GP appointments in England were conducted by telephone. What happens during these consultations is largely unknown. AIM: To test the feasibility of collecting recorded GP telephone consultations with linked data and consent for reuse. DESIGN & SETTING: Retrospective observational study in seven practices in South West England collected from April 2023 to June 2024. METHOD: Adults who had a telephone consultation at practices that routinely recorded calls were invited to consent to retrieval of the audio, a 4-month electronic health record (EHR) extract, and a post-consultation questionnaire. Practice-level consent rates were analysed with regression. RESULTS: Of 28 clinicians recruited, 19 GPs contributed retrievable, usable recordings with consent for research re-use. Of 2,053 patients invited, 123 consented (6.0%), with lower consent in more deprived practices (IMD deciles 1-2 versus 9-10: OR 0.22, 95% CI 0.09-0.54). Of 101 recordings retrieved, 96 were usable and 91 had consent for research re-use. Of these, 86 were linked to EHR data and 89 to patient questionnaires. Mean consultation duration was 7 minutes 13 seconds, and typing was audible in 63/91 (69%). Overall, 161 problems were discussed, averaging 1.77 per consultation. Most patients were happy their consultation was by telephone (96/117, 82%), although most reported usually preferring face-to-face appointments (68/115, 59%). CONCLUSION: It is feasible to assemble a reusable archive of GP telephone consultations with linked data. However, recruitment was low using retrospective remote consent. Future work should test alternative recruitment approaches and consider oversampling at practices serving deprived populations.