OBJECTIVES: Treatment of carious primary molars is always indicated, especially on young children; however, there are no clear guidelines that precisely explain the best treatment approach for Class II carious molars with marginal breakdown (International Caries Detection and Assessment System [ICDAS] 5). The objective of this prospective observational clinical study was to assess the efficacy of three restorative techniques in treating ICDAS 5 Class II lesions in primary molars: compomer fillings (CF), preformed metal crowns (PMC), and pulpotomy and conventional preformed metal crowns (PMC+P). The secondary goal was to evaluate the impact of some cofactors on the course of treatment. METHOD AND MATERIALS: Overall, 92 children (female, n = 50, 54.3%; male, n = 42, 45.7%) aged 2 to 9 years old (mean age = 5.9 ± 1.9 years) with 166 treated teeth were included. The average number of decayed, missing, or filled teeth (d3mft) of the whole sample was 8.0 ± 3.4. The distribution of the sample according to type of treatment was CF = 53 (31.9%), PMC = 64 (38.6%), and PMC+P = 49 (29.5%). Paired t test, nonparametric Friedman ANOVA test, and decision tree analysis were used as the basis for the statistics. RESULTS: After 12 months, data from 75.8% (72/95) treated patients, corresponding to 62.0% (103/166) of the treated teeth (CF = 42/53, 79.2%; PMC = 38/64, 59.4%; PMC+P = 23/49, 46.9%) were available for analysis. The mean patients age was 6.8 ± 1.8 years; 32 (47.1%) boys and 36 (52.9%) girls. The mean d3mft of the remaining sample was 7.8 ± 3.35. PMC and PMC+P arms showed the highest success rates (> 91%) as compared to the CF arm, which showed the lowest success rates (61.9%), with 9/42 teeth of the CF group (21.4%) presenting with minor failures, and 7/42 teeth (16.7%) with major failures (P < .0001). CONCLUSION: According to the decision tree analysis, PMC and PMC+P had a success rate of 99%, whereas CF had a success rate of only 69%. Some cofactors (treatment decision, Approximal Plaque Index, and tooth number) had a higher impact on the decision tree analysis than others (age, dmfs, and dmft values), especially when the treatment selection was CF. In future studies it is necessary to examine the impact of other cofactors on the outcomes of conventional fillings using a larger sample size.