Images captured using light-induced fluorescence systems are generally analysed using proprietary software. The purpose of this study was to determine if such images could be scored visually and to compare these data with those metrics produced by the software. A total of 171 lesions were selected from a pool of images to provide a range of lesions which were reported as having remained static, increased or decreased in fluorescence using the QLF analysis software. The baseline and 6-month images were then assessed side by side on a computer screen by 10 examiners who were asked to rate the lesions to determine if the lesion had become better or worse, or had stayed the same. There was generally poor correlation between clinical visual image assessments and all QLF analysis outcomes for all examiners. The agreement amongst the visual image assessments for the 10 examiners compared to the average score ranged from kappa 0.22 to 0.59 and the rank correlations from -0.01 to 0.73. This study suggests that the visual assessment of lesion images by the examiners in this study was based upon different characteristics of lesion change than those utilised by the QLF analysis software. A clearer understanding of lesion characteristics that are indicative of positive and negative changes may be required before this technology can be exploited to its full potential. Copyright © 2007 S. Karger AG.