Resource use and conservation objectives can conflict where protected species and commercially fished species occur together. We tested the potential for a previously overfished deep-sea shark (Centrophorus zeehaani) with very low biological productivity to recover in a network of three reserves. An individual-based simulation model was developed and applied using reproduction and movement data. Without any reserves or reductions in fishing mortality the population will never recover and will fail in 27.9 ± 6.1 years. With three reserves, recovery from 8% of initial numbers to a target of 20% would take 63.1 ± 3.1 years with the network in place. Length of the female cycle, natural mortality, and density dependent female dispersal are uncertain, but could delay recovery time by an additional 16.5, 98.3, or 61.9 years, respectively. Effectiveness of the reserve network was particularly sensitive to location and patchiness of the population. Doubling the size of a reserve where C. zeehaani are abundant would reduce recovery time by 12.5 years; halving it would increase recovery time by 12.0 years. Future re-opening of orange roughy fishing in waters deeper than 750 m would delay recovery of C. zeehaani by 45.9 years. The methods developed here can be used to explore co-management options for target species and protected species in fisheries.