Purpose: (1) To compare SEM quantitative marginal analysis data with the depth of penetration (DP) of the three most commonly used tracers for microleakage in Class II fillings in vitro; (2) based on the obtained results to calculate the discriminatory power of a sample size of 12. Materials and Methods: Standardized large cavities (mesially 1 mm above the CEJ, distally 1 mm below the CEJ, intercuspal distance 70%) were prepared into 36 extracted caries-free first mandibular molars and filled with Tetric EvoCeram/AdheSE, the resin being applied in horizontal layers 2 mm thick. Each increment was light cured for 20 s (1200 mW/cm2). Finishing was performed with fine diamond burs and disks. All teeth were subjected to occlusal loading (1,200,000 cycles, 49 N/1.7 Hz) and simultaneous thermocycling (3000 cycles at 5°C/55°C). The percentage of continuous margin of the cervical dentin and enamel was evaluated on replicas using SEM. The teeth were subjected to tracer penetration with either 0.5% basic fuchsin (24 h, 37°C), 2% methylene blue (24 h, 37°C) or 50% silver nitrate solution (4 h, 37°C, followed by 8-h exposure to a photodeveloping solution and overnight fluorescent light). The teeth were mesiodistally sectioned twice. The depth of tracer penetration was measured with a stereomicroscope and averaged for each site. Due to data inhomogeneity and abnormal distribution, both SEM and DP data were transformed. Sample size calculations were performed based on standard deviation and statistical error estimates. Results: For the dentin margin, there was an acceptable correlation between SEM data and both fuchsin penetration (Pearson: -0.74, p 0.01) and silver nitrate penetration (Pearson: -0.79, p 0.01), but not between methylene blue and SEM data. For enamel margins, no significant correlation between SEM and DP data was found for the three tracers. There was statistically no significant difference in tracer penetration between the three tracers (Kruskal Wallis, p > 0.05). For all three tracers, statistically significantly higher penetration occurred at the dentin than at the enamel margin (Wilcoxon, p 0.05). A sample size of 12 makes it possible to discriminate between materials only when they differ in tracer penetration in the range of 1 mm for enamel and 2 mm for dentinal margins. Conclusion: Tracer penetration with fuchsin or silver nitrate showed a moderate correlation with SEM quantitative marginal analysis data at dentinal margins, but not at enamel margins. Keywords: microleakage, marginal analysis, restoration, tracer, dentin adhesives