Objectives: The purpose of this prospective longitudinal cohort study was to 1) test the hypothesis that there is no significant difference in early failure of restorations that are either repaired or replaced, and 2) identify treatment characteristics associated with early failures for these two treatment modalities. Methods: Practitioners in the National Dental PBRN (NationalDentalPBRN.org) enrolled patients who had defective restorations that needed repair or replacement in permanent teeth. Investigators completed a data collection form at the time of the restoration repair or new restoration placement, and evaluated the restorations at annual follow up. This report comes from the first visit after 6-18 months. Failure was defined as the restoration being rated not acceptable and requiring either repair, replacement, endodontic treatment, or extraction. Previous root canal or extraction was also recorded. Results: 191 practitioners treated 6,018 defective restorations, of which complete data were available on 5,761. Repaired restorations (n= 1,458) resulted in 1,303 rated as acceptable and 155 failures (11%; 129 repaired or replaced, 12 received endodontics, and 14 were extracted). Replaced restorations (n= 4,303) resulted in 4,062 rated as acceptable and 241 failures (6%; 182 repaired or replaced, 40 received endodontics, and 19 were extracted). The rate of restoration failure was significantly higher for repaired (11%) than for replaced restorations (6%, p < .001), however, there was no difference in the frequencies of endodontics and tooth extraction between repaired and replaced restorations. Restoration failure was not associated with type of tooth, upper or lower arch, and number of surfaces involved. Conclusion: The results of this prospective cohort study showed that repaired restorations had a significantly higher failure rate during the first year of observation than restorations that were replaced.