In 2004, an American Society of Clinical Oncology (ASCO) technology assessment on adjuvant use of aromatase inhibitors (AIs) was updated to indicate that optimal adjuvant endocrine therapy for postmenopausal women with receptor-positive breast cancer should include an AI, either as up-front therapy or as sequential therapy after tamoxifen. This recommendation was on the basis of improved disease-free survival (DFS) observed with AIs, given that no trial at that time had demonstrated improvement in overall survival. This recommendation and the publicity relating to trials such as the Arimidex, Tamoxifen, Alone or in Combination (ATAC) trial have had substantial impact worldwide on the endocrine treatment of early breast cancer. On the basis of the recent published update of the ATAC trial with a median 8 or more years of follow-up, and the report at the San Antonio Breast Cancer Meeting in December 2007, advertisements proclaim that the long-term data demonstrate continuing effectiveness for anastrozole over tamoxifen through treatment completion and beyond (for example, see the advertisement in the March 10, 2008 issue of this journal; Vol 26, No. 8). The goals of therapy for any stage or type of cancer are to improve the duration and/or quality of survival. If we are to switch from an inexpensive and well-established treatment such as tamoxifen to a more expensive and newer alternative (an AI), then the onus of the companies and investigators supporting that switch must be to prove that the new treatment is more effective in one or both of those key end points. The ATAC study is a large and well-designed trial that compares up-front therapy using anastrozole with that using tamoxifen for 5 years in postmenopausal women with hormone-sensitive early breast cancer. Results of the trial continue to show a significant improvement in DFS in favor of anastrozole (hazard ratio [HR] 0.85), but they do not show improved overall survival (HR 1.0). A related study, Breast International Group 1-98 (BIG 1-98), which is comparing up-front letrozole therapy with up-front tamoxifen therapy for 5 years, also shows improved DFS in favor of letrozole (HR 0.82), but differences in overall survival are not significant after a relatively short median follow-up of 51 months (HR 0.91). There are several reasons why improvement in DFS may not translate into improved overall survival for women participating in trials of adjuvant therapy for breast cancer. First, the definition of DFS varies among trials evaluating AIs, but these studies included local recurrence and new primary breast cancer (including preinvasive disease in the ATAC but not the BIG 1-98 trial), which have a minimal effect on overall survival. Second, an initial delay to relapse for women receiving up-front anastrozole might be balanced by longer survival after relapse for those who receive tamoxifen. Third, an improvement in cancer-specific survival might be counterbalanced by an increase in deaths as a result of other causes; indeed, there is a trend for such an effect in the ATAC trial, given that the 8-year results show a nonsignificant reduction in deaths after recurrence for the anastrozole arm counterbalanced by an increase in deaths without recurrence. Timetrends in the three publications reporting outcomes from the ATAC trial at median follow-up of 33, 68, and 100 months show an increasing excess in noncancer deaths for women receiving anastrozole as compared with tamoxifen, whereas the deficit in deaths after cancer recurrence for women receiving anastrozole has remained constant in the last two reports (Table 1); this raises concerns about the long-term serious toxicity of anastrozole. There is strong evidence that DFS is a good surrogate for overall survival in trials of adjuvant therapy for colorectal cancer, but support for surrogacy of DFS in adjuvant trials for breast cancer is less solid. In some trials, an early large improvement in DFS has translated into a smaller later improvement in overall survival; however, smaller differences in DFS at short follow-up, as observed in trials comparing different endocrine therapies, may not be associated with subsequent differences in overall survival. If overall survival is similar after adjuvant treatment with an AI or tamoxifen, up-front use of an AI might be preferred if it is better tolerated than tamoxifen and leads to better quality of life. Some would argue that quality of life will be better if there is a delay of recurrence of breast cancer, but that depends on the toxicity of treatment. The ATAC investigators have published specific analyses of quality of life and of adverse effects of treatment. They evaluated quality of life using the Functional Assessment of Cancer Treatment– Breast (FACT-B) and its endocrine subscale, and reported no overall differences between the arms of the study. The abstract of their article describing adverse effects indicates that treatment-related adverse effects occurred significantly less often with anastrozole than with tamoxifen, as did treatment-related serious adverse events and adverse events leading to withdrawal. However, these statements seem inconsistent with data provided in the article. For example, there JOURNAL OF CLINICAL ONCOLOGY COMMENTS AND CONTROVERSIES VOLUME 27 NUMBER 6 FEBRUARY 2