Hypoxia (low dissolved oxygen) is a mounting concern for aquatic ecosystems as its prevalence increases with rising anthropogenic nutrient inputs. Hypoxia is most commonly defined as 2.0 mg l–1 of dissolved oxygen, although this level varies widely across studies and agency regulations. Such definitions may be too conservative, as ecologically-relevant non-lethal effects (e.g. consumption and growth) of hypoxia on important aquatic species, such as fish, often occur at oxygen levels much higher than 2.0 mg l–1. In addition, many mechanisms that regulate hypoxia tolerance in fish have been proposed, including temperature, habitat, location in the water column, and body size, but there is ongoing debate over which mechanisms are most important. Using a structured meta-analysis of published studies, we showed consistent, significant negative effects on fish growth and consumption below 4.5 mg l–1. While the total amount of variation explained was generally low, below 4.5 mg l–1 of dissolved oxygen, phylogenetic relationships accounted for most of the explained variation in fish growth. Ecological factors including body size, location in the water column (pelagic, demersal, or benthopelagic), habitat (freshwater, marine, or diadromous), and temperature explained very little of the effect of hypoxia on fish growth and explained only a moderate level of variation in consumption. Our results suggest a dramatically higher threshold for sub-lethal effects of hypoxia on fish than oxygen levels generally set for regulation purposes, and provide little support for accepted ecological mechanisms thought to influence hypoxia tolerance.