Genetic rescue is increasingly viewed as a promising but underutilized conservation strategy to mitigate inbreeding depression and restore genetic diversity. Yet, empirical evidence supporting its long-term efficacy is limited to a few generations. Here, we conducted an experiment with Drosophila melanogaster to test the long-term efficiency of genetic rescue in reducing population extinction risk. A recently captured wild population was maintained in the laboratory with a large census size (the base population, BP). Thirty smaller populations of N = 50 individuals (N50) were founded from the BP and maintained for 31 generations. Subsequently, three sets of 54 lines with N = 8 individuals each, were founded from these N50 populations and maintained for 33 generations. One set served as a control (non-rescued lines), while the other two sets were rescued using a single male introduced in two consecutive generations either from the BP (rescued-BP) or from the N50 (rescued-N50) populations. An analysis of pupae productivity at the time of the foundation of the lines showed substantial purging of the inbreeding load in the N50 populations. No significant differences in pupae productivity were found between the non-rescued lines and rescued-BP lines at generation 9. However, extinction rates were lower for the latter. Whole-genome sequencing of 12 individuals from a non-rescued line and 12 from a rescued-BP line revealed fewer deleterious allele copies and a lower inbreeding in the rescued one. Our results highlight the effectiveness of genetic rescue in reducing the extinction rate by efficiently selecting advantageous alleles responsible for hybrid vigour, purging of the introduced genetic load, and introducing adaptive potential.