The objective of this study was to assess the feasibility of selecting for functional longevity in rabbits, defined as an ability to delay involuntary culling. Functional longevity was measured as the total number of AI performed after the first kindling. Breeding values were estimated using a discrete survival model. Male parents were selected on the basis of their progeny test results, and the efficiency of selection was estimated in the second generation, as was the correlated response on reproduction traits. A total of 48 males were progeny tested, based on the longevity of 10 daughters bred in 2 different farms. Based on their estimated genetic merit, 5 "high longevity" (HL) and 5 "low longevity" (LL) males were selected divergently and produced a new generation (5 bucks/sires and 10 daughters/bucks). A difference in longevity (+0.75 AI, i.e., 32 d) was observed between the 2 lines. In farm 1, the differences were mainly due to culling (26% in the LL line vs. 14% in the HL line) whereas mortality was similar in the 2 lines. In farm 2, mortality and culling were both higher in the LL line than in the HL line (33 vs. 15% and 19 vs. 7%, respectively). There was no difference between the 2 lines in terms of the reproduction traits recorded for each kindling. Nevertheless, because of the difference in the litter number between the 2 lines, the sum of young rabbits born alive per doe over her lifetime and the sum of young rabbits weaned per doe were higher in the HL line (+5 kits; P < 0.01). Selection for functional longevity using survival analysis is feasible for modifying lifetime reproduction traits.