Modern concrete dams have devastated fish populations world-wide. However, dams vary greatly in how they are engineered and operated, and thus pose a range of threats to riverine fauna. Understanding the differences in the impacts of dams is critical for setting conservation priorities. We used a modified BACI (before-after-control-impact) sampling design as a means to quantify the effects of dams on spring/summer chinook salmon in two watersheds (Snake and Upper Columbia Rivers) of the Columbia River Basin, USA. The construction of four dams in the Columbia River Basin from 1966 to 1975 allowed us to test the hypothesis that the presence of these dams does not affect the abundance, survival and population growth of chinook salmon. In both the Snake and Upper Columbia Rivers, there was a significant decline from the period before dams were constructed (1959-65) to the period after dams were constructed (1980-90). In the Upper Columbia River, declines in productivity or population performance (measured as recruits per spawner or Ricker function residuals) were greater than in the control region. On the other hand, patterns of fish productivity in the Snake River were similar to those seen in the control region. The disparity between fates of Upper Columbia and Snake River populations points to the differences between regions in current efforts to reduce fish mortality associated with dams. Our analysis suggests that dams in the Upper Columbia River, but not Snake River, are a potential force preventing recovery of endangered salmon populations.