Abstract Captive breeding can be a useful tool for the management of threatened species, but such programs often begin with a bottleneck which may lower reproductive fitness by increasing inbreeding and allowing deleterious alleles to increase in frequency by reducing effective population size. Dryococelus australis, the Lord Howe Island stick insect, is a critically endangered insect that has been bred at the Melbourne Zoo in Melbourne, Australia, since shortly after the species was rediscovered in 2001. We analysed a long‐term dataset of phenotypic measurements from this captive breeding program to determine whether reproductive fitness has declined in captivity. We found a clear signal of decline in direct (egg hatching success) and indirect (egg and nymph size) fitness indicators over time. We compared a captive line descended from the original program founders to a new hybrid line that has recent wild ancestry. The hybrid line showed an immediate improvement in egg hatch rate, suggesting the potential for genetic rescue. Egg and nymph size did not show as dramatic a change, although this line had only existed for just over a single generation at the time of data collection. These results argue strongly for the regular sourcing of new genetic material from the wild D. australis population to counteract fitness declines in captivity. Aside from benefiting the captive population, this would also improve the prospects of establishing new wild populations in the future, using individuals sourced from captivity.