Aromatase inhibitors (AIs) are approved for use in both early- and advanced-stage breast cancer in postmenopausal women. Although the currently approved "third-generation" AIs all powerfully inhibit estrogen synthesis, they may be subdivided into steroidal and nonsteroidal inhibitors, which interact with the aromatase enzyme differently. Nonsteroidal AIs bind noncovalently and reversibly to the aromatase protein, whereas steroidal AIs may bind covalently and irreversibly to the aromatase enzyme. The steroidal AI exemestane may exert androgenic effects, but the clinical relevance of this has yet to be determined. Switching between steroidal and nonsteroidal AIs produces modest additional clinical benefits, suggesting partial noncrossresistance between the classes of inhibitor. In these circumstances, the response rates to the second AI have generally been low; additional research is needed regarding the optimal sequence of AIs. To date, clinical studies suggest that combining an estrogen-receptor blocker with a nonsteroidal AI does not improve efficacy, while combination with a steroidal AI has not been evaluated. Results from head-to-head trials comparing steroidal and nonsteroidal AIs will determine whether meaningful clinical differences in efficacy or adverse events exist between the classes of AI. This review summarizes the available evidence regarding known differences and evaluates their potential clinical impact.