It is generally assumed that shivering and exercise are analogous processes. This assumption most often finds expression in mathematical models of human thermoregulation where substrate utilization during exercise is used as the model for shivering. This may be erroneous given the different patterns of muscle recruitment, temperatures, and levels of stress associated with each activity. The present experiment compared, for the first time in humans, substrate utilization during shivering and exercise.The protocol was approved by the Ethics Committee of Simon Fraser University. Eight male subjects undertook two, 2 h experiments. In the first they were immersed, naked, to the xiphoid process in 20°C water. Low (L, average oxygen consumption (Vo2) 0-49 1 min-), low-moderate (LM, average Vo2 0-61 lmin-) and moderate (M, average Vo02 0-9 1 min-) intensity shivering was elicited from the subjects by cooling and stirring the water, and spraying the neck and shoulders. During the second experiment, the V02 measured during shivering was reproduced in each subject by bicycle ergometry. During both experiments V02 was recorded every 10 s using on-line sampling and a computer-based data acquisition system. The rates of hepatic glucose output (HGO) and utilization (Rd) were determined using radioimmunoassay following primed constant infusions of sterile, pyrogen-free [3-3H] glucose. Arterialized blood samples were taken every 10 min from a vein in the dorsum of the hand from which plasma free fatty acid (FFA) and ,-hydroxybutyrate (,-HBA) concentrations were determined. The Wilcoxon signed ranks test was used to analyse the data.None of the variables recorded differed between the two conditions at rest. The shivering and exercise V02 of the subjects were closely matched, the highest average (2 S.D.)variation observed being 0f02 (0 03) 1 min-'. HGO (,rIE) has been shown to be reduced in humans by procedures designed to augment peripheral chemosensitivity, such as acute hypoxaemia (Griffiths et al. 1986) and chronic metabolic acidaemia (Oren et al. 1982). However, as these conditions are not normally encountered during exercise at sea level, we wished to examine whether an endogenously induced, acute metabolic acidaemia (which is commonly encountered) might have a similar effect.Six healthy subjects performed an incremental cycle ergometer test to estimate the lactate threshold (6L) and peak V02. Subsequently (on different days), they completed two tests, each consisting of two consecutive, 6 min exercise bouts separated by 6 min at '0' Watts. In both instances, the second (or 'test') bout was at a work rate equivalent to 90 % of 6L; the initial (or 'conditioning') bout was also sub-9L (90% 6L) in one case, and supra-OL (at a work rate halfway between OL and peak fo2) in the other (i.e. to generate an acute metabolic acidaemia). Pulmonary gas exchange variables were derived breath-by-breath from continuous monitoring of respiratory volumes (turbine) and gas concentrations (mass spectrometer). During minute 6 of each exerc...