Purpose: To report retro-prospective data on the prevalence of overall implant failure in a large number of edentulous patients treated at one referral clinic over a 30-year period and to analyze possible associations between implant failure and basic clinical variables. Materials and Methods: Altogether, 24,781 implants were consecutively placed in 4,585 edentulous arches between 1986 and 2015. All implant failures identified at the clinic during follow-up were consecutively recorded, and a multivariate logistic regression analysis was performed to identify possible associations between implant failure and different clinical factors. Results: Altogether, 1,333, 688, and 249 treated arches were followed up for 15, 20, and 25 years, respectively. Cumulative survival rates (CSR) for the treated arches were 86.2% and 83.8% after 15 and 25 years, respectively. Most patients lost only one implant each (58%). Loss of all implants was reported in 68 arches, with total failure rates of 1.9% and 2.2% after 15 and 25 years, respectively. The strongest associations with increased risk for implant failure were maxilla (hazard ratio [HR] 4.76; 95% confidence interval [CI] 3.70 to 6.25) and implant surface (HR 2.38; 95% CI 1.59 to 3.57). Age at surgery, implant surgeon, calendar year of surgery, and time of follow-up also showed significant associations with risk of implant failure (P .05). A completely steady-state level in implant survival was not observed, but few implants were lost up to the last years of follow-up. Conclusion: There is a higher risk for implant failure in the maxilla compared to the mandible. Risk is reduced when using implants with a moderately rough surface. The highest risk for failure was observed during the first year. This was followed by a reduced failure rate, which never reached a steady-state level.