Oxidative respiration is strongly temperature driven. However, in woody stems, efflux of CO(2) to the atmosphere (E (A)), commonly used to estimate the rate of respiration (R (S)), and stem temperature (T (st)) have often been poorly correlated, which we hypothesized was due to transport of respired CO(2) in xylem sap, especially under high rates of sap flow (f (s)). To test this, we measured E (A), T (st), f (s) and xylem sap CO(2) concentrations ([CO(2)*]) in 3-year-old Populus deltoides trees under different weather conditions (sunny and rainy days) in autumn. We also calculated R (S) by mass balance as the sum of both outward and internal CO(2) fluxes and hypothesized that R (S) would correlate better with T (st) than E (A). We found that E (A) sometimes correlated well with T (st), but not on sunny mornings and afternoons or on rainy days. When the temperature effect on E (A) was accounted for, a clear positive relationship between E (A) and xylem [CO(2)*] was found. [CO(2)*] varied diurnally and increased substantially at night and during periods of rain. Changes in [CO(2)*] were related to changes in f (s) but not T (st). We conclude that changes in both respiration and internal CO(2) transport altered E (A). The dominant component flux of R (S) was E (A). However, on a 24-h basis, the internal transport flux represented 9-18% and 3-7% of R (S) on sunny and rainy days, respectively, indicating that the contribution of stem respiration to forest C balance may be larger than previously estimated based on E (A) measurements. Unexpectedly, the relationship between R (S) and T (st) was sometimes weak in two of the three trees. We conclude that in addition to temperature, other factors such as water deficits or substrate availability exert control on the rate of stem respiration so that simple temperature functions are not sufficient to predict stem respiration.