Canopy and soil CO2 concentration and isotopic measurements were conducted in a slash pine forest during a progressive drought in the southeastern United States (1) to determine and compare variations in δ13C of foliage (δCf), soil (δCs), and ecosystem-respired CO2 (δCr) and (2) to evaluate the usefulness of a two end-member oxygen isotope ratio (δ18O of CO2) approach to partition nighttime ecosystem respiration into soil and plant components at different heights within the canopy. The δCf was enriched by 2.2 ¿ 0.3 (standard error) ? during the extreme drought in May relative to September when precipitation was above normal. The enrichment in δCf exceeded changes in δCr and δCs for the same time period (1.6 ¿ 0.5 and 1.0 ¿ 0.4?, respectively). Lower variations in the 13C of soil-respired CO2 relative to the variations of the autotrophic component can buffer changes in δCr, which integrates the 13C signature of canopy and soil respired CO2. Application of a two end-member model to canopy δ18O-CO2 indicated that soil CO2 contribution to nighttime CO2 buildup within the canopy decreased from 100% at the soil surface to 0% within the canopy in September. In May, during the extreme drought period, soil respiration rate was 2.7 times lower than the rate in September despite similar soil temperatures (23¿ in May versus 19¿C in September). Soil and aboveground respiration equally contributed to the nocturnal CO2 buildup within the canopy in May. Our model was limited in that it produced an upper limit value for soil respiration and neglected respiration by woody tissue.