The objective of this study is to determine whether eliminating uncertainty in diagnosis improves accuracy in teleradiology for remote trauma management and to assess the correlation between the probable reason for the mistake and the level of confidence. An emergency physician interpreted 2133 radiological series transmitted from a community hospital to an emergency department using videoconferencing equipment. He marked his level of confidence as 1 = sure positive; 2 = fairly sure positive; 3 = unsure; 4 = fairly sure negative; 5 = sure negative. The emergency physician also entered free-text comments, e.g., feeling tired, to enable a better understanding of what was happening. The accuracy, sensitivity, and specificity were 97.3% (98.6%), 94.48% (96.72%), and 98.63% (99.42%), respectively. Figures in brackets represent the 93.5% of cases in which the level of confidence was graded as sure. Boredom and hurry were the main reasons identified for mistakes. Reducing uncertainty in diagnosis may improve the accuracy of screen interpretation. This first series must be reproducible by others before any general use of such systems can be made for primary diagnosis. American College of Radiology standards must be met.