Studies of time-invariant matrix metapopulation models indicate that population growth rate is more sensitive to the vital rates of individuals in “good” patches than in “bad” patches. This suggests that management and conservation efforts should focus on the best patches, while inferior patches should be given lower priority. Most populations, however, experience a variable environment. For example, the vital rates of nearshore benthic invertebrates are influenced by disturbances such as waves, predation, weather, and anthropogenic activity. We examined the effects of stochastic disturbance on the sensitivity of metapopulation growth using a two-patch matrix metapopulation model based on the life cycle of the softshell clam, Mya arenaria. Individuals in the two patches differ in fecundity, so that in the absence of dispersal one patch is of high quality (log λ_s > 0) and the other of low quality (log λ_s < 0). Disturbance reduced fecundity at equal frequency and with the same intensity in both patches. We show the stochastic metapopulation growth rate and quasi-extinction probability are sometimes more sensitive to the fecundity of individuals in the bad patch, rather than in the good patch. Our analysis uses combinations of disturbance intensity, frequency, and covariance typical of scenarios encountered by M. arenaria. However, it can be expanded to include additional patches, multiple stages, stochastic dispersal, and complex demography.