All of four fish species used in 174 laboratory experiments with a “channeled avoidance tank” avoided some low oxygen concentrations, the degree of avoidance generally decreasing with increasing concentration. Avoidance indices computed were based on numbers of entries into two experimental channels with reduced oxygen concentrations and two control channels, on numbers of crossings of a transverse line located well inside each channel, and on numbers of fish observed in the channels at 60‐second intervals. Juvenile chinook salmon showed marked avoidance of oxygen concentrations near 1.5, 3.0, and 4.5 mg./1. in summer at high temperatures. A decrease of the avoidance observed in the fall is ascribable to lower temperatures. The chinook salmon showed little avoidance of concentrations near 4.5 mg./1. in the fall, and no avoidance of concentrations near 6.0 mg./1. at any time. At summer temperatures juvenile coho salmon showed some avoidance of all the above oxygen concentrations, including 6 mg./1., but their behavior was erratic and their avoidance of concentrations near 4.5 mg./1. and less was not as pronounced as that of chinook salmon at corresponding temperatures. Largemouth bass and bluegill markedly avoided concentrations near 1.5 mg./1., but showed little or no avoidance of the higher concentrations, only the bass showing any avoidance of concentrations near 4.5 mg./1. The nature and significance of the observed avoidance reactions are not quite clear. The prompt avoidance by some fish of oxygen concentrations well above those known to be lethal cannot be ascribed entirely to mere stimulation or increase of activity due to oxygen deficiency. It may not be assumed, however, that oxygen concentrations avoided under the experimental conditions are usually avoided likewise under natural conditions.