A full analysis of optimal fisheries investment strategies must take into account high levels of uncertainty in future fishery returns, as well as irreversibility of investment in specialized, nonmalleable fishing fleets. A stochastic optimization model is analyzed using dynamic programming to determine optimal policy functions for both fleet investment and fish stock management within an uncertain environment. The resulting policies are qualitatively similar to those found in the corresponding deterministic case, but quantitative differences can be substantial. Simulation results show that optimal fleet capacity should be expected to fluctuate over a fairly wide range, induced by stochastic variations in the biomass. However, the performance of a linear-cost risk-neutral fishery is fairly insensitive to variations in investment and escapement policies around their optimum levels, so that economic optimization is "forgiving" within this context. A framework of balancing upside and downside investment risks is used here to explain the roles of several fishery parameters in relation to optimal investment under uncertainty. In particular, the intrinsic growth rate of the resource and the ratio of unit capital costs to unit operating costs are found to be key parameters in determining whether investment should be higher or lower under uncertainty.