Animals given substances with thyroid activity show a reduced resistance to anoxia (Duran, 1920), and Smith, Emmens, and Parkes (1947) have used this effect in a method of thyroid assay based on the survival times of mice in closed jars. We have used it for comparing the activities of thyroxine and some of its derivatives. We find it relatively simple to carry out; the results it gave have been confirmed in clinical trials, both with sodium thyroxine and with the N-formyl derivative (Hart and Maclagan, 1950).We were particularly interested in two questions: has solubility an important effect on the activity of thyroxine and has thyroxine enhanced activity in the dried glandular extract ? For this purpose a number of preparations of synthetic thyroxine and its derivatives were prepared (Clayton and Hems, 1949) and examined for thyroid activity. This paper records the results obtained biologically. PROCEDURE Throughout the whole of the experiment the mice were kept in a thermostatically controlled incubator room at a temperature of 25.50 ± 1°. For each experiment use was made of male fawn mice of the GFF strain, weighing between 18 and 24 grammes, distributed into groups of twenty animals to give uniform body weight representation. Usually six groups, at ascending dose levels, were employed, three for the standard and three for the test preparations; a further twenty mice were kept as controls.The doses were administered on three alternate days, every animal in a group receiving the same dose irrespective of body weight and the volume of the solution injected being the same for all groups. All compounds for assay were either suspended in 0.9 per cent sodium chloride solution or dissolved in 0.01 M-sodium carbonate solution containing 0.9 per cent sodium chloride.On the second day after the final injection the groups of ten mice were sealed in glass jars and the time of sealing was recorded. We used 32-oz. wide neck, screwcapped clear bottles, with soft paraffin smeared on the neck to ensure a complete seal. The volumes of the jars ranged from 980 to 1,020 ml., and they were randomly distributed. The time of survival was recorded for each individual mouse, the endpoint being the last visible respiration, which is almost invariably preceded by marked terminal convulsions.