A prospective survey was conducted to assess physician responsibility in melanoma prognosis. Consecutive patients with primary melanoma were interviewed and examined using a standardized questionnaire. Main outcome measures were medical components of the delay before tumor resection and tumor thickness. Of 590 melanomas, 29.1% were coincidentally detected by physicians and their tumor depth was lower than in melanomas detected by patients (p < 0.001). Physician sensitivity for melanoma diagnosis was evaluated at 86%. Median time intervals to propose resection and to perform removal of melanoma were short: 0 (mean 103) and 7 (mean 68) days, respectively. Melanomas were managed in an inappropriate way in 14.2% of cases. Location on acral areas and absence of pigmentation were associated with longer medical delays and more frequent inappropriate medical attitudes. Melanomas located on hardly visible areas were less frequently detected by physicians than those on visible areas. Medical delays were shorter, doctor's attitude was more frequently appropriate, and melanoma thickness was lower (p < 0.001) when the patient visited a dermatologist (54.7%) than when he or she visited a general practitioner (33.4%). Our study shows that doctor responsibility accounts for only a small part of the total delay before melanoma removal. However, systematic total examination and better training of doctors, especially about unusual forms of melanoma, could still improve melanoma detection.