The third-generation aromatase inhibitors (AIs) are gradually displacing tamoxifen as the preferred adjuvant endocrine treatment for hormone-receptor-positive early breast cancer in postmenopausal women, having demonstrated superior efficacy in clinical trials. However, for the AIs to gain widespread acceptance in the adjuvant setting, good long-term tolerability must also be demonstrated, particularly as many women with early breast cancer can now expect to live for over a decade after initial diagnosis. Tamoxifen has been widely used in this setting for over 30 years, and the side effects associated with its long-term use are well documented. Many adverse events that occur with tamoxifen are predictable consequences of its antiestrogenic actions, including hot flushes and mood disturbances. However, tamoxifen also has estrogenic properties in some tissues, which are associated with desirable and undesirable effects. Of particular importance, tamoxifen can cause unwanted gynecological events, including an increased risk of developing endometrial cancer, and thromboembolic disease. Conversely, tamoxifen protects against postmenopausal bone loss, modestly lowers cholesterol levels and may protect against cardiac disease. As the number of women treated with AIs increases, long-term safety data relating to these agents will gradually accumulate. Safety data are currently available from early adjuvant trials comparing an AI with tamoxifen during the first 5 years after surgery (sequential and substitution strategies), and from the extended adjuvant setting, comparing letrozole with placebo after completion of 5 years of tamoxifen therapy (the MA.17 trial). In early adjuvant studies, the use of tamoxifen as a comparator can complicate the analysis of safety data, particularly in tissues where tamoxifen has beneficial, estrogenic effects. Studies in the early adjuvant setting have shown that AIs are generally well tolerated and are associated with a lower incidence of vaginal bleeding, thromboembolic disease and endometrial cancer than tamoxifen. Results from these studies have suggested that AIs are associated with musculoskeletal side effects, including bone loss, osteoporosis and fractures, cardiovascular disease and hypercholesterolemia. However, analysis of safety data from MA.17, with a placebo control, showed no evidence of adverse effects of letrozole on the cardiovascular system or lipid profiles. When the data from early and extended adjuvant studies are considered together, it can be concluded that the increased incidences of hypercholesterolemia and cardiovascular disease seen in patients taking an AI probably reflect the lack of the protective effects of tamoxifen in these patients rather than a detrimental effect of the AI. Bone loss is a predictable consequence of the near-complete elimination of circulating estrogen achieved by third-generation AIs in postmenopausal women. In the MA.17 trial, although more women on letrozole reported new, self-diagnosed osteoporosis than those on placebo, the number of clinical fractures did not differ significantly between the two treatment arms, however, further follow-up is needed. Furthermore, ongoing studies are evaluating the role of bisphosphonates and other agents to better characterize and potentially ameliorate AI-associated bone loss in postmenopausal women. The adverse events associated with AI use are predictable and manageable. On the basis of current data, the tolerability of AIs in the adjuvant setting appears as good as that of tamoxifen, and some serious adverse events associated with tamoxifen use are avoided. Further studies and longer follow-up from current trials will help to determine in more detail the long-term effects of this class of drugs.