Aromatase inhibitors (AIs) are major treatment options for the management of patients with breast cancer. The drugs are effective and response rates can be high. However, resistance, either primary or acquired during treatment, may occur. Optimal clinical management requires accurate predictors of response to identify those tumours, which are most likely to respond (so sparing patients with resistant tumours needless side-effects of ineffective therapy). Currently, oestrogen receptor (ER) status is the only factor used routinely to select for treatment with aromatase inhibitors, but a substantial proportion of ER-positive tumours fails treatment. There is, therefore, an urgent need to identify additional markers by which accurately to predict clinical response on an individual basis. Whilst other markers (such as progesterone receptors or HER2) have some predictive powers, individually they have limited utility for routine use. The hope is that discovery strategies based on genome-wide searches will identify novel markers that can be used as predictive indices. Molecular phenotyping of individual tumours could then be used to decide not only which patients should be treated with AIs but whether AIs should be used alone or in combination or in sequence with other targeted agents in order that clinical benefits are maximized. This chapter will focus on utility of routinely assessed biomarkers, multi-component indices and gene signatures and outline the future perspectives in studies aimed to AI response prediction.