Captive breeding programs assume that fitness decline under domestication selection proceeds at a roughly constant rate per generation, allowing managers to estimate safe captive durations. Reanalysis of Araki et al.'s (2007, 2009) steelhead trout (Oncorhynchus mykiss) data reveals that the decline accelerates: per-generation wild-fitness loss increases from 20% (G0→G1) to 54% (G1→G2), a 2.7× rate increase. This acceleration is predicted by Christie et al.'s (2012) finding that captive fitness and wild fitness are negatively correlated — alleles favored in captivity actively degrade wild performance. Each generation's adaptation to captive conditions deepens the mismatch with the wild environment, producing a positive feedback loop: domestication selection is autocatalytic. Inbreeding accounts for only 1–4% of the observed decline (Christie et al., 2012), ruling out genetic drift as the primary mechanism. The Biggins et al. (1998) black-footed ferret data provide an independent demonstration that niche provision depth — not genetics — drives the fitness cost: cage-reared and pen-reared ferrets from the same breeding program show 2% vs. 20% post-release survival, a 10-fold difference with genetics controlled. These findings suggest that conservation breeding protocols should minimize niche provision (enriched semi-natural rearing rather than sterile captive environments) and should plan for accelerating, not constant, fitness decline per captive generation.