Fishes (Gillichthys mirabilis) having different temperature histories were subjected to KCN, to boiled sea water, and to temperatures high or low enough to be lethal. Urethane, an anesthetic, was used in a limited number of experiments. The time of death (or of anesthetization) was recorded for each of the fishes (nearly 3,000). Counts of respiration were made in a considerable number of cases.Death in both KCN and in boiled water was speedy in proportion to the temperature, the reciprocals of the times of death forming an approximate logarithmic series when plotted against temperature. The temperatures used were 10°, 15°, 20°, 25° and 30°.In the case of KCN, a low correlation was found to exist between the duration of previous acclimatization to high or low temperatures and the degree of resistance to KCN at those temperatures. The abrupt decrease of resistance (increase in metabolic rate), resulting from transfer to a higher temperature, was followed by a slight increase of resistance which continued for several days. On the other hand, the abrupt increase of resistance (decrease in metabolic rate), resulting from transfer to a lower temperature, was followed by a slight decrease which continued for several days.In boiled water, this compensatory trend was only shown after transfer to higher temperatures.When fishes which had been acclimatized to high (30°) and low (10°) temperatures were tested in KCN, boiled water or urethane at 20°, there was a consistent reversal in the resistance of the respective lots, the 30° fishes now being much more resistant than the 10° ones. These relations continued for several days after transfer to the intermediate temperature.Death in de-oxygenated sea water was more than 60 per cent slower, at the same temperatures, than in KCN solution.Acclimatization at 30°, even for brief periods, resulted in a marked resistance to the lethal effects of heat. This result was already conspicuous, after a half hour at 30°, and increased, at a diminishing rate up to 10 days or more.Return to a common temperature of fishes having "cold" and "warm" histories diminished, but did not annul, the effects of previous acclimatization upon heat resistance. Pronounced differences were noted in one experiment, after 23 days at the intermediate temperature. Interesting relations were pointed out between the period of conditioning and the persistence of the effects.Acclimatization at 10° increased resistance to lethal cold (1.0°—).Fishes kept for alternating periods at 10°, 20°, and 30° (even when the 10° periods were twice as long as the 30° ones) were more resistant to lethal heat than fishes kept continuously at 20°. Such fishes were, however, less resistant to lethal cold than the controls.The rate of the visible respiratory movements was much greater at 20° and 30° than at 10°. There was no certain difference, however, between the rates at the first two temperatures.Transfer of fishes of "cold" and "warm" history to an intermediate temperature resulted in the same reversal in relative respiratory rates as was found in the case of resistance to KCN, etc.The size of the fish appears to have little or no influence upon its resistance to lethal heat or to cyanide poisoning. On the other hand, it was found that smaller fishes had a more rapid respiratory rhythm, and succumbed more rapidly in boiled water than larger ones.Acclimatization to higher or lower temperatures brings about two classes of effects which seem to be largely distinct from one another. One of these concerns increased resistance to the harmful effects of heat and cold, when these are used as lethal agents. The second of these concerns changes in respiratory metabolism, and therefore in resistance to oxygen lack, including KCN poisoning. Reasons are given for believing that the former effect is not dependent upon the latter.