The aim of this clinical study was to present potential sources of errors influencing the analysis of fluorescence images by using Quantitative Light-induced Fluorescence (QLF). Fluorescence images of 1835 lingual/ buccal surfaces and 517 occlusal fissures of thirty four 15-yr-old adolescents were captured after professional tooth-cleaning four times during a longitudinal study. Images were taken with a CCD camera with the mirror version in a dental office without any ambient light under standardised conditions. The computer program (Inspektor QLF 2.00) was used to display, store and analyse the images. QLF images were analysed by two trained examiners. Average fluorescence loss (DF) of the lesion, the area (A) of the lesion (mm2) and DQ (DF*A) were determined. Based on a large amount of analysed fluorescence images following confounding factors were compiled: 1. Despite of professional tooth-cleaning performed before QLF recordings red fluorescing areas caused by residual plaque/ calculus may be limited longitudinal measurements. 2. Different brightness of fluorescence images observed in several clinical examinations can pretend to remineralisation of lesions. 3. The analytical stage of the QLF method is limited if lesions extended to the gingival margin or to more than one surface. 4. Surfaces attacked by hypoplasia and developmental disorders have to be differentiate clinically from carious lesions before the QLF assessment. 5. Angular errors during image capturing can not be adjusted or eliminated completely by the time-consuming analytical process. An efficient control of all confounding factors is an important prerequisite in longitudinal studies. The new Video Repositioning System may be able to eliminate some sources of errors in future studies. Keywords: QLF, clinical, diagnostics, optical methods of caries diagnosis, fluoresence technique