Inaccurate coding of patients' Underlying Cause of Death (UCOD) has constrained cause-specific survival estimates for colon and rectal cancers. Using California data from the Accuracy of Cancer Mortality study, we compared the cancer site data from the California Cancer Registry (CCR) with UCODs reported on death certificates and reclassified the UCODs based on cancer registry data when they disagreed. We then calculated 1-, 3-, 5-, and 10-year cause-specific survival for colon and rectal cancers separately, before and after the reclassification. Records from 26 312 colon and 10 687 rectal cancer patients were examined. UCOD records disagreed with CCR records for 700 (6%) of 11 404 colon cancer deaths and with 1958 (39%) of 5011 rectal cancer deaths, and 82% of the misclassified rectal cancer deaths were coded as colon cancer deaths in the UCOD. Reclassification decreased cause-specific survival for both colon and rectal cancers, but the impact was more pronounced for rectal cancer (eg, 5-year cause-specific survival of colon cancer decreased by 2.8% and of rectal cancer decreased by 20.0% relative to previous estimates; absolute rates changed from 65.4% to 63.6%, and 81.2% to 64.9%, respectively, after reclassification). Interchangeable use of the terms colon cancer and colorectal cancer is likely to be one of the reasons for UCOD misclassification. Educational measures could improve the accuracy of UCOD for colon and rectal cancer deaths.