Background There is an inherent tension between allowing trainees in general practice (GP) to feel comfortable to report and learn from errors in a blame-free environment while still assuring high-quality and safe patient care. Unfortunately, little is known about the types and potential severity of incidents that may confront GP trainees. Furthermore, incident reporting by resident trainees is hindered by their concern that such transparency might result in more negative performance evaluations. Objective To explore the number and nature of incidents that were reported by GP trainees and to determine whether there were differences between the reporters and non-reporters based on their performance evaluations. Design Prospective cohort study. Methods Confidential and voluntary incident reporting was implemented in GP vocational training of the University Medical Center Utrecht, the Netherlands. Seventy-nine GP trainees were asked to report incidents over 6&emsp14;months. Mixed methods were used to analyse the data. Results 24 trainees reported a total of 44 incidents. 23 incidents concerned the work process and 17 concerned problems with diagnosis or therapy. Three-quarters (34/44) of incidents were determined to be not specifically related to the inexperience of the GP trainees. While actual patient harm was determined to be minimal or absent in two-thirds of incidents (29/44), the potential for moderate, major, or catastrophic harm was 89% (39/44). Trainees performing best on their performance assessment in the domain of clinical expertise reported incidents more often (43% vs 18%, p<0.03) than those who performed at a lower level. Conclusions GP trainees rated highly by their faculty voluntarily reported incidents in the delivery of clinical care when given a safe, blame-free, and confidential reporting process. Most incidents were not found to be directly related to the inexperience of the trainee, but were caused by failing organisational processes in the healthcare delivery system. Moreover, the ...