PURPOSE OF REVIEW: Asthma attacks are frequent in children with asthma and can lead to significant adverse outcomes including time off school, hospital admission and death. Identifying children at risk of an asthma attack affords the opportunity to prevent attacks and improve outcomes. RECENT FINDINGS: Clinical features, patient behaviours and characteristics, physiological factors, environmental data and biomarkers are all associated with asthma attacks and can be used in asthma exacerbation prediction models. Recent studies have better characterized children at risk of an attack: history of a severe exacerbation in the previous 12 months, poor adherence and current poor control are important features which should alert healthcare professionals to the need for remedial action. There is increasing interest in the use of biomarkers. A number of novel biomarkers, including patterns of volatile organic compounds in exhaled breath, show promise. Biomarkers are likely to be of greatest utility if measured frequently and combined with other measures. To date, most prediction models are based on epidemiological data and population-based risk. The use of digital technology affords the opportunity to collect large amounts of real-time data, including clinical and physiological measurements and combine these with environmental data to develop personal risk scores. These developments need to be matched by changes in clinical guidelines away from a focus on current asthma control and stepwise escalation in drug therapy towards inclusion of personal risk scores and tailored management strategies including nonpharmacological approaches. SUMMARY: There have been significant steps towards personalized prediction models of asthma attacks. The utility of such models needs to be tested in the ability not only to predict attacks but also to reduce them.