On three separate occasions, five zebra finches ( Taenopygia guttata) were injected intraperitoneally with 0.2 ml 0.29 M NaH(13)CO(3)solution and placed immediately into respirometry chambers to explore the link between (13)C elimination and both O(2) consumption (VO(2)) and CO(2) production (VCO(2)). Isotope elimination was best modelled by a mono-exponential decay. The elimination rate (k(c)) of the (13)C isotope in breath was compared to VO(2) (ml O(2)/min) and VCO(2) (ml CO(2)/min) over sequential 5-min time intervals following administration of the isotope. Elimination rates measured 15-20 min after injection gave the closest relationships to VO(2) ( r(2) =0.82) and VCO(2) ( r(2)=0.63). Adding the bicarbonate pool size (N(c)) into the prediction did not improve the fit. A second group of birds ( n=11) were flown for 2 min (three times in ten birds and twice in one) between 15 min and 20 min following an injection of 0.2 ml of the same NaH(13)CO(3) solution. Breath samples, collected before and after flight, were used to calculate k(c) over the flight period, which was converted to VO(2) and VCO(2) using the equation generated in the validation experiment for the corresponding time period. The energy expenditure (watts) during flight was calculated from these values using the average RQ measured during flight of 0.79. The average flight cost measured using the bicarbonate technique was 2.24+/-0.11 W (mean+/-SE). This average flight cost did not differ significantly from predictions generated by an allometric equation formulated by Masman and Klaassen (1987 Auk 104:603-616). It was however substantially higher than the predictions based on the aerodynamic model of Pennycuick (1989 Oxford University Press), which assumes an efficiency of 0.23 for flight. The flight efficiency in these birds was 0.11 using this model. Flight cost was not related to within-individual variation [general linear model (GLM) F(1,31)=1.16, P=0.29] or across-individual variations in body mass (GLM F(1,31)=0.26, P=0.61), wingspan (regression F(1,10)=0.01, P=0.94) or wing loading (regression F(1, 31)=0.001, P=0.99) in this sample of birds.