Microcalorimetry was used to study the effects of graded hypoxia and anoxia on two species of insects that differ in their tolerance of anoxia. Locusts (Locusta migratoria) can survive an atmosphere of pure nitrogen for not more than 4 h (at room temperature), whereas hawk moths (Manduca sexta) can recover from more than 24 h of anoxia. To produce graded hypoxia, air and pure nitrogen were mixed and this mixture was passed through the cells of a twin calorimeter equipped with circulation cells. A gas flow containing 2% or more of oxygen had no significant effect on behaviour (as observed in parallel experiments using transparent cells) or heat flow rate. If oxygen content was reduced to 1% or less the effects of oxygen lack became conspicuous; at a "critical oxygen concentration" (between 2 and 1% oxygen) the animals became agitated; they hyperventilated and showed escape movements, which were followed by a loss of body posture and complete immobility within a few minutes. This behaviour was reflected by a distinct peak in heat flow rate followed by a precipitous decrease in heat flow below the normoxic rate. During graded hypoxia, the heat flow rate approached a new value which was correlated with the degree of hypoxia. Under strict anoxia, 5.3 +/- 1.1% and 3.6 +/- 1.8% of the normoxic heat flow rates were reached by Locusta and Manduca, respectively. Thus the two insect species reacted similarly with respect to behaviour and metabolic rate (as indicated by heat flow) to both graded hypoxia and anoxia. A striking difference between the species was seen, however, in the rate of heat flow during recovery from hypoxia or anoxia. Readmittance of air after an anoxic or hypoxic interval led to a rapid increase in heat flow above the normoxic rate, but the amount of "excess heat production" during recovery was much higher in Locusta than in Manduca. The difference in the amount of "excess heat production" during posthypoxic recovery could not be accounted for by differences in behaviour (muscular activity) or by the effect of ambient temperature.