Objectives: The purpose of this prospective longitudinal cohort study was to test if the restorative material used for repair or replacement was associated with the longevity of the restorations. Methods: Practitioners in the National Dental PBRN (NationalDentalPBRN.org) enrolled patients who had defective restorations that needed either repair or replacement in permanent teeth. Investigators completed a baseline data collection form at the time of the restoration repair or new restoration placement, and then evaluated the restorations with annual follow up and subsequent reports. 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. Separate analyses were performed for restorations that were repaired and replaced. Results: A total of 191 practitioners evaluated 6,018 treated defective restorations, of which treatment and outcome data were available on 5,850. A total of 1,483 restorations were repaired (25%) and 4,367 restorations were replaced (75%). The mean follow-up period was 11.5 (SD=3.7) months. A total of 396 restorations failed, for a failure rate of 7%. When the restoration was repaired, the overall failure rate was significantly different by material used for the repair [p = .019; amalgam (12%), direct RBC (9%), indirect restorations (16%)], but not for the material in the original restoration and partially left in the repair treatment [p = .645; amalgam (11%), direct RBC (10%), indirect restorations (11%)]. When the restoration was replaced, the overall failure rate was also significantly different by replacement material [p <.001; amalgam (9%), direct RBC (5%), indirect restorations (5%)], but not for the material in the original restoration. Conclusion: The choice of either amalgam or RBC restorations for restorations that were repaired or replaced was predictive of restoration failure.