Continuous automatic chamber and eddy covariance (EC) measurements were made at an old aspen forest (SOA) located at southern boreal treeline in Saskatchewan, Canada to examine the temporal variability in soil (Rs), tree bole (Rb), and ecosystem respiration (RE) during 2001. Climatic conditions were significantly warmer and drier than the 30-year climate normal, resulting in lower RE and an unprecedented increase in net ecosystem productivity (NEP). In the 7-year record (1994, and 1996–2001) of CO2 exchange at SOA, the year 2001 showed the greatest carbon gain (300 g C m−2 year). Scaled chamber measurements (1315 g C m−2 per year) were 37% larger than the EC estimate of RE (961 g C m−2 per year). The difference between the scaled chambers and the EC estimate was reduced to 20% after correcting for lack of energy balance closure. Annual RE was approximately 170 g C m−2 per year lower than the average of the previous 6 years. Annual estimates of microbial-heterotrophic (Rh) (510 g C m−2 per year) and autotrophic respiration (Ra) (805 g C m−2 per year), based on chamber measurements, were used to help validate the EC estimate of RE. Ra represented 61% of the total chamber respiration. This fraction was used to partition RE into Ra and Rh to calculate net primary production (NPP). The values of NEP (300 g C m−2 per year) and NPP (675 g C m−2 per year) were more characteristic of temperate forests. The NPP/Pg ratio of 0.54 was within the range of recently published values using biometric techniques and supports that the annual ecosystem respiration budget and its partitioning was well constrained. We recognize, however, that this ratio will vary interannually depending on climatic conditions. In 7 years of annual EC CO2 flux measurements at SOA, this study provides the first evidence that drought can lead to a transient increase in CO2 sequestration resulting from a reduction of RE.