SUMMARYExperiments have been made with rainbow trout in solutions of ammonium chloride and zinc sulphate to examine the empirical rule that a mixture of two poisons,PandQ, should be at the threshold concentration for acute toxicity whenPS/PT + QS/QT= I, where the suffixSstands for the concentration in solution andTfor the threshold concentration of a poison when tested by itself. The most detailed experiment, with a hard dilution water in which the dissolved‐oxygen concentration was at the air‐saturation value, gave results within 4% of those expected from the hypothesis. Agreement was not so close in a soft dilution water, nor in a hard dilution water containing a reduced concentration of dissolved oxygen, but in these two cases the threshold concentrations for the mixtures were within 26 and 17% of the expected value of unity. In every case the rule under examination provided a better description of the data than models based on the hypothesis of independent joint act