The net ecosystem exchange of CO2 between forests and the atmosphere, measured by eddy covariance, is the small difference between two large fluxes of photosynthesis and respiration. Chamber measurements of soil surface CO2 efflux (Fs), wood respiration (Fw) and foliage respiration (Ff) help identify the contributions of these individual components to net ecosystem exchange. Models developed from the chamber data also provide independent estimates of respiration costs. We measured CO2 efflux with chambers periodically in 1996–97 in a ponderosa pine forest in Oregon, scaled these measurements to the ecosystem, and computed annual totals for respiration by component. We also compared estimated half-hourly ecosystem respiration at night (Fnc) with eddy covariance measurements. Mean foliage respiration normalized to 10 °C was 0.20 μmol m–2 (hemi-leaf surface area) s–1, and reached a maximum of 0.24 μmol m–2 HSA s–1between days 162 and 208. Mean wood respiration normalized to 10 °C was 5.9 μmol m–3sapwood s–1, with slightly higher rates in mid-summer, when growth occurs. There was no significant difference (P > 0.10) between wood respiration of young (45 years) and old trees (250 years). Soil surface respiration normalized to 10 °C ranged from 0.7 to 3.0 μmol m–2 (ground) s–1 from days 23 to 329, with the lowest rates in winter and highest rates in late spring. Annual CO2 flux from soil surface, foliage and wood was 683, 157, and 54 g C m–2 y–1, with soil fluxes responsible for 76% of ecosystem respiration. The ratio of net primary production to gross primary production was 0.45, consistent with values for conifer sites in Oregon and Australia, but higher than values reported for boreal coniferous forests. Below-ground carbon allocation (root turnover and respiration, estimated as Fs– litterfall carbon) consumed 61% of GPP; high ratios such as this are typical of sites with more water and nutrient constraints. The chamber estimates were moderately correlated with change in CO2 storage in the canopy (Fstor) on calm nights (friction velocity u* < 0.25 m s–1; R2 = 0.60); Fstor was not significantly different from summed chamber estimates. On windy nights (u* > 0.25 m s–1), the sum of turbulent flux measured above the canopy by eddy covariance and Fstor was only weakly correlated with summed chamber estimates (R2 = 0.14); the eddy covariance estimates were lower than chamber estimates by 50%.