To develop and validate a software program which automatically detects gaps of cylindrical fillings in bovine dentin by using an optical sensor (FRT MicroProf), evaluate 8 different adhesive systems with the defined parameters, and compare the results with that of SEM quantitative marginal analysis. An optical measuring head with strong wavelength dependency on its focal length splits white light into its colors. The surface height can be measured from the wavelength of the reflected light (resolution: x/y 1 to 2 µm, z 10 nm). A profile of the surface is taken by scanning the measuring point radially from the center of the filling over the restoration margins. An algorithm detects gaps and calculates their depths and widths. Eight adhesive systems (AdheSE, Clearfil SE, Excite, iBond, Optibond Solo Plus, Prompt L-Pop, Touch and Bond, Xeno III) were applied in cylindrical cavities (n = 8, Ø = 4 mm, depth 2.5 mm, C-factor 3.5) in bovine dentin together with Tetric Ceram. In the control group, no adhesive system was used. After 24 h of storage in distilled water at 37°C, the surface was polished with 4000-grit abrasive paper, and then epoxy resin replicas were produced. The evaluation criteria were as follows: the percentage of continuous margin along the entire margin (SEM, 500X) and the percentage of gap-free profiles to total profiles (FRT MicroProf), the latter being evaluated by using 100, 300 and 1000 profiles per sample. Two groups (Touch and Bond, Optibond Solo Plus) with 16 samples were selected to evaluate reproducibility and the occurrence of voids adjacent to the restoration margins. For each adhesive, a paired t-test was applied to evaluate the differences between both test methods for each adhesive. A linear regression analysis procedure was used to evaluate the overall agreement between the two test methods. The mean deviation between three consecutive measurements was 2.75%. Ten percent of all profiles showed voids adjacent to the restoration margins; 92% of the voids were excluded by setting the threshold value of the gap width at 20 µm. In less than 1% of the profiles, the concurrent presence of gap and void resulted in the nondetection of gaps. It was found that 100 profiles per sample are sufficient to characterize a sample as compared to SEM analysis. Except for the control group, the paired t-test did not reveal any statistically significant difference between the two test methods for either adhesive (p 0.05). Regression analysis for all 72 data points yielded R² = 0.72 (p 0.0001). The results of the quantitative marginal analysis by means of the FRT optical sensor are comparable to those obtained by SEM. The advantages of this method include (1) saving of operator time, (2) independence of operator's experience and bias, and (3) additional information about the gap width and depth. Palabras clave: marginal analysis, restoration, optical sensor, adhesive systems