RATIONALE: Severe asthma is a real public health problem because of its consequences on patients and its economic cost. Nevertheless, few epidemiological studies have focussed on severe asthma. The aim of this state of the art review is to describe the criteria used to estimate asthma severity and to assess the prevalence of severe asthma from population-based epidemiological studies. OBSERVATIONS: Asthma severity has been defined in various ways in epidemiology, with criteria based on symptoms, hospitalisation, ventilatory function, discomfort caused by asthma attacks, and treatment. A high variability in the prevalence of severe asthma is observed, resulting mostly from the lack of standardisation in the phenotypes used, but also from a real geographical heterogeneity. Nevertheless, the prevalence of severe asthma may be reasonably estimated between 1 and 3% of the general population, both in children and adults. CONCLUSION: Data from the literature show a lack of standardisation of the severe asthma phenotype. Further epidemiological studies are needed to understand the geographical variability in severe asthma prevalence.