Under the conditions of both an increased red cell affinity for O2 at a constant rate of O2 delivery (arterial O2 content × flow) and a decrease in the rate of O2 delivery induced by hypoxic hypoxia at constant blood flow, we have obtained a linear relationship between the partial pressure of O2 in the muscle venous effluent (Pv,O2) and O2 uptake (V˙O2). The relationship is described by the equation V˙O2 = Da × Pv,O2 + V˙O2,conv where Da is the apparent O2 diffusion capacity and V˙O2,conv is O2 delivery-limited V˙O2, and Da × Pv,O2 represents the O2 diffusion-limited V˙O2 (V˙O2,diff). From these observations, we propose the hypothesis that V˙O2 consists of two additive values, V˙O2,conv and V˙O2,diff. The mechanism underlying the reduction in V˙O2 that is induced by reducing O2 delivery to markedly below the V˙O2,conv value has only been investigated using a model based on the single compartment of diffusion-limited V˙O2, and has not been investigated in terms of this additive V˙O2 model. The single compartment analysis appears to overestimate the role of O2 diffusion in limiting the reduction of V˙O2 that occurs in response to a decrease in O2 diffusion capacity, as reflected by the V˙O2/Pv,O2 ratio. To gain better insight into the mechanism involved, we altered the rate of O2 delivery by changing arterial PO2 from normoxia (with inhalation of air) to hypoxia (by inhalation of 10-11 % O2) and blood flow (with high and low flow rates (n = 7 for both groups), and very low and ischaemic flow rates (n = 4 for both groups)) in pump-perfused dog gastrocnemius preparations during tetanic isometric contractions at 1 Hz. As rates of O2 delivery were reduced from 23.2 to 10.9 ml min-1 (100 g)-1, significant decreases in Pv,O2 and V˙O2 were observed (P < 0.05). From the data of Pv,O2 and V˙O2 values within this range of O2 delivery rates, we obtained the regression equation V˙O2 = 0.22 × Pv,O2 + 8.14 (r = 0.58). From the equation, the intercept of the V˙O2-axis was significantly different from zero (P < 0.05), in accordance with the observation that the V˙O2/Pv,O2 ratio (ml min-1 (100 g)-1 Torr-1) increased from 0.54 to 1.35 (P < 0.05). However, at extremely low rates of O2 delivery (5.6 and 7.3 ml min-1 (100 g)-1 the V˙O2/Pv,O2 ratio was 1.51 and 2.80 (P < 0.05), respectively. This indicates a break in the linear V˙O2-Pv,O2 relationship as the rate of O2 delivery was reduced to below the V˙O2,conv value of the V˙O2-axis intercept. These results suggest that the reduction in V˙O2 caused by extreme reductions in the rate of O2 delivery is not attributable to a reduction in O2 diffusion capacity, as expected from the V˙O2/Pv,O2 ratio, but to a reduction in the O2 delivery-limited V˙O2 component, as evaluated by the V˙O2-axis intercept of the linear V˙O2-Pv,O2 relationship. Experimental Physiology (2002) 87.1, 53-61.