A new inventory of CO emissions in China is presented for the year 2001. This inventory improves and updates the a priori CO emission inventory prepared in support of NASA's TRACE-P mission in the spring of 2001. Analysis of CO observations using chemical transport models in inverse and forward modes suggested that China's emissions were underestimated by about 50%. We have reexamined the source characteristics and conclude that emissions from cement kilns, brick kilns, and the iron and steel industry were underestimated. Our new estimate for China's CO emissions in 2001 is 157 Tg, 36% higher than the TRACE-P estimate for the year 2000 of 116 Tg. Bottom-up and modeled emission estimates are now in good agreement, which represents a major success story for the TRACE-P mission. The new inventory has been gridded at 30 min ¿ 30 min resolution and tested with the CFORS/STEM-2K1 model, considerably improving the correlation between model predictions and observations (bias reduced from 27% to 9%). Propagation-of-error estimates in the new inventory yield an uncertainty of ¿68% (95% confidence intervals), lower than the TRACE-P value of ¿156%; however, the good agreement with results from inverse and forward models implies a greater level of confidence than this. The largest remaining uncertainties concern (1) characterization of open vegetation burning, which cannot be resolved without new field studies; (2) emission factors for small combustion devices, for which emissions testing is urgently needed; and (3) residential fuel consumption, which may require a reassessment of China's official statistics.