Author Posting. © American Geophysical Union, 2008. This article is posted here by permission of American Geophysical Union for personal use, not for redistribution. The definitive version was published in Journal of Geophysical Research 113 (2008): C03019, doi:10.1029/2007JC004153. ; Estimates of temporal trends in oceanic anthropogenic carbon dioxide (CO2) rely on the ability of empirical methods to remove the large natural variability of the ocean carbon system. A coupled carbon-climate model is used to evaluate these empirical methods. Both the ΔC* and multiple linear regression (MLR) techniques reproduce the predicted increase in dissolved inorganic carbon for the majority of the ocean and have similar average percent errors for decadal differences (24.1% and 25.5%, respectively). However, this study identifies several regions where these methods may introduce errors. Of particular note are mode and deep water formation regions, where changes in air-sea disequilibrium and structure in the MLR residuals introduce errors. These results have significant implications for decadal repeat hydrography programs, indicating the need for subannual sampling in certain regions of the oceans in order to better constrain the natural variability in the system and to robustly estimate the intrusion of anthropogenic CO2. ; We would like to acknowledge funding from NSF (OCE02-23869), NCAR, the WHOI Ocean Climate Institute, a Linden Earth Systems Graduate Fellowship (MIT), and a National Defense Science and Engineering Graduate Fellowship. NCAR is sponsored by the National Science Foundation. R.W. is supported by the Office of Oceanic and Atmospheric Research at NOAA.