Asthma presents in various clinical forms and levels of severity and has a complex pathophysiology. Unbiased clustering was initially performed on clinical features, but the addition of biomarkers such as sputum and blood cellular profiles has led to the description of several phenotypes and enabled the prediction of responses to targeted therapies. Clusters of severe asthma include those on high-dose corticosteroid treatment, often with both inhaled and oral treatment, associated with severe airflow obstruction. Concordance between symptoms and sputum eosinophilia is observed in an eosinophilic inflammation-predominant group with few symptoms and late-onset disease who have a high prevalence of rhinosinusitis, aspirin sensitivity, and exacerbations. Sputum or blood eosinophilia is also a biomarker that can predict therapeutic responses to antibody-based treatments to block the effects of the T-helper-2 cytokine, interleukin-5. Low T-helper-2 expression predicts poor therapeutic response to inhaled corticosteroid therapy. Much less is known about 'non-eosinophilic' or non-T-helper-2 asthma. Clustering on transcriptomic and/or proteomic data is leading to definition of molecular phenotypes and potential mechanisms. The definition of endotypes and biomarkers of disease and therapeutic responses will pave the way towards personalized medicine and healthcare for asthma. For the clinician, these will translate into useful tools and means for managing patients with asthma, particularly severe asthma. Corresponding author : Kian Fan Chung, f.chung@imperial.ac.uk