Widespread losses of production and conservation values make large-scale ecosystem restoration increasingly urgent. Ecological restoration by means of biomanipulation, i.e., by fishing out planktivores to reduce the predation pressure on herbivorous zooplankton, has proved to be an effective tool in restoring degraded lakes and coastal ecosystems. Whether biomanipulation may prove a useful restoration method in open and structurally complex marine ecosystems is, however, still unknown. To promote a recovery of the collapsed stock of Eastern Baltic cod (Gadus morhua), large-scale biomanipulation of sprat (Sprattus sprattus), the main planktivore in the Baltic Sea, has been suggested as a possible management approach. We study the effect of biomanipulation on sprat using a statistical food-web model, which integrates internal interactions between the main fish species of the Central Baltic Sea, with external forcing through commercial fishing, zooplankton, and climate. By running multiple, stochastic simulations of reductions in sprat spawning stock biomass (SSB) only minor increases in cod SSB were detected, none of which brought the cod significantly above ecologically safe levels. On the contrary, reductions in cod fishing mortality and/or improved climatic conditions would promote a significant recovery of the stock. By this we demonstrate that an ecosystem-scale biomanipulation, with the main focus of reinstating the dominance of cod in the Baltic Sea may likely be ecologically ineffective, operationally difficult, and costly. We argue that reducing exploitation pressure on Eastern Baltic cod to ecologically sound levels is a far more appealing management strategy in promoting a long-term recovery and a sustainable fishery of the stock.