Aims: To test the assumption that professional recall of consultation decisions is valid and more accurate than patient recall of consultation decisions. Methods: One hundred and thirty-four consultations between diabetes specialist nurses and diabetes specialist dietitians in an adult out-patient diabetes service were audiotaped. Patients and professionals were asked to recall the treatment decisions made immediately after the consultation. Patient participants were also asked to complete the Health Care Climate Questionnaire (HCC). Recalled decisions, by patient and professional participants, were then compared with those extracted from the audio tapes, and with each other. Results: The mean duration of consultations was 27 min. Patients recalled a mean of 2.5 (sd 1.4) decisions per consultation, and professionals a mean of 3.2 (sd 1.6) decisions per consultation. A mean of 2.2 (sd 1.1, range 0-4) decisions per consultation were identified on the audiotapes. Patients recalled a mean of 2.3 (sd 1.4, range 0-6) decisions per consultations that could not be found on the tapes, with professionals recalling a mean of 1.7 (sd 1.2, range 0-6) decisions per consultation that could not be found on the tape. More autonomy, as measured by the HCCQ, was correlated with better professional recall (r = 0.17; P < 0.05). Conclusions: Both patients and professionals have poor recall of decisions made in diabetes out-patient consultations. Although the mean professional recall is marginally better than that of the patients, they recall a vast number of unmade decisions and the implications of these being recorded in patients' notes is substantial.