Abstract The number of cormorants has rapidly increased in the northernmost Baltic Sea. In 2018, 50 km × 50 km ICES catch rectangle 55H1 had 3140 breeding pairs. To estimate the predation effect of cormorants on perch populations, we Passive Integrated Tags tagged 1977 perch and 9.9% of tags were found. The median instantaneous cormorant-induced mortality during the breeding time, with consumption by non-breeding individuals, was estimated at 0.23 and at 0.35 during the whole residing period. We estimated with a yeild-per-recruit model that the long-term maximum loss of perch yield of tagged sub-population would be at 80% probability interval 32–67%, and when extended to the entire 55H1, 10–33%, respectively. The cormorants’ share of the >2-year-old perch biomass and production would be 8%, while that of other natural mortality would be 63% and that of fishing 29% in 55H1. The yield-per-recruit-results should be interpreted as an estimate of maximum cormorant effect because the dependence of predation rate on prey density was not accounted for, and density-dependence of growth, mortality, and reproduction of perch could partly compensate the loss. The results indicate that high density of cormorants can reduce the perch stocks and catches locally.