The immediate and delayed effects of embryonic exposure to low levels of polynuclear aromatic hydrocarbons (PAHs) have been shown to reduce survival to maturity by 50% in exposed pink salmon populations. This suggests that chronically exposed populations could be extirpated over relatively few generations, but the effect of density dependence on extirpation rate is unknown. This study examines the interaction of PAH effects and randomly varying density dependence on a simulated population. The simulation derives from 70 years of observations made on a single pink salmon population and toxicity studies involving a hatchery population in the same watershed. Results from simulations involving exposure of 100% of the population to effects consistent with an aqueous PAH concentration of 18 nL/L indicate an 80% decrease in population productivity and an 11% probability of extinction after 35 generations. In contrast, population growth rate declined by only 5%. Further decreases in survival relative to that of observed PAH effects rapidly increase the probability of extinction. Data from these simulations demonstrate that, at low levels of exposure, density dependence can compensate for reduced population size and buffer the population against extinction. However, if equilibrium size is depressed sufficiently, random environmental variation overcomes the buffering effect of density dependence and extinction probability increases. These data demonstrate that extinction probability and population size are more sensitive measures of population effects than growth rate for wild populations regulated by density dependence.