Over 1 billion USD are devoted annually to rehabilitating freshwater habitats to improve survival for the recovery of endangered salmon populations. Mitigation often requires the creation of new habitat (e.g. habitat compensation) to offset population losses from human activities, however compensation schemes are rarely evaluated. Anadromous Pacific salmon are ecologically, culturally, and economically important in the US and Canada, and face numerous threats from climate change, over-harvesting, and degradation of freshwater habitats. Here we used a matrix population model of coho salmon (Oncorhynchus kisutch) to determine the amount of habitat compensation needed to offset mortality (2-20% per year) caused by a range of development activities. We simulated chronic mortality to three different life stages (egg, parr, smolt/adult), individually and in combination, to mimic impacts from development, and evaluated if the number of smolts produced from constructed side-channels demographically offset losses. We show that under ideal conditions, the typical size of a constructed side-channel in the Pacific Northwest (PNW) (3405 m2) is sufficient to compensate for only relatively low levels of chronic mortality to either the parr or smolt/adult stages (2-7% per year), but populations do not recover if mortality is >10% per year. When we assumed lower productivity (e.g.; 25th percentile), or imposed mortality at multiple life stages, we found that constructed channels would need to be larger (0.2-4.5 times) than if we assumed mean productivity or as compared to the typical size built in the PNW, respectively, to maintain population sizes.. We conclude that habitat compensation has the potential to mitigate chronic mortality to early life stages, but that current practices are likely not sufficient when we incorporate more realistic assumptions about productivity of constructed side-channels and cumulative effects of anthropogenic disturbances on multiple life stages.