Hypoxia is an environmental stressor that affects abundance, biomass, diversity and ecosystem function of benthic assemblages worldwide, yet its collective impact at individual, population, and ecosystem levels has rarely been investigated. We examined the effects of hypoxia on the biomass-dominant clam, Macoma balthica, in the York and Rappahannock Rivers (Chesapeake Bay, USA). We (i) surveyed the M. balthica populations in both rivers in 2003 and 2004, (ii) determined the effects of low DO on M. balthica fecundity in a laboratory experiment, and (iii) employed a predator-exclusion field experiment to establish the effects of hypoxia and prey density on predation upon M. balthica. The resultant data were used to parameterize a matrix model, which was analyzed to define potential effects of hypoxia at the population level. In both rivers, hypoxia decreased individual clam growth and caused local extinction of populations. Hypoxia reduced egg production of M. balthica by 40% and increased protein investment per egg. In the predator-exclusion field experiment, hypoxia magnified predation rates three-fold and altered the functional response of predators to M. balthica from a stabilizing type III functional response to a destabilizing type II functional response. In a density-independent matrix model, hypoxia resulted in coupled source-sink metapopulation dynamics, with hypoxic areas acting as black-hole sinks. Increases in the spatial and temporal extent of hypoxia caused the populations to decline toward extinction. In a second model that incorporated density-dependence, under mild hypoxic conditions trophic transfer from M. balthica to predators increased, but at increased spatial or temporal extent of hypoxia trophic transfer decreased. The major decline in trophic transfer to predators under severe hypoxia resulted from diversion of M. balthica biomass into the microbial loop. Our model predicted that there are multiple stable states for M. balthica populations (high and very low densities), such that the saddle point (threshold at which the population switches from one state to the other) increased and resilience decreased with the spatial extent of hypoxia. We underscore how effects of a stressor at the individual level can combine to have substantial population and ecosystem-level effects.