Current understanding of risk factors for asthma attacks in children is based on studies of small but well-characterised populations or pharmaco-epidemiology studies of large but poorly characterised populations. We describe an observational study of factors linked to future asthma attacks in large number of well-characterised children. From two UK primary care databases (Clinical Practice Research Datalink and Optimum Patient Care research Database), a cohort of children was identified with asthma aged 5-12 years and where data were available for ≥2 consecutive years. In the "baseline" year, predictors included treatment step, number of attacks, blood eosinophil count, peak flow and obesity. In the "outcome" year the number of attacks was determined and related to predictors. There were 3776 children, of whom 525 (14%) had ≥1 attack in the outcome year. The odds ratio (OR) for one attack was 3.7 (95% Confidence Interval (CI) 2.9, 4.8) for children with 1 attack in the baseline year and increased to 7.7 (95% CI 5.6, 10.7) for those with ≥2 attacks, relative to no attacks. Higher treatment step, younger age, lower respiratory tract infections, reduced peak flow and eosinophil count >400/μL were also associated with small increases in OR for an asthma attack during the outcome year. In this large population, several factors were associated with a future asthma attack, but a past history of attacks was most strongly associated with future attacks. Interventions aimed at reducing the risk for asthma attacks could use primary care records to identify children at risk for asthma attacks.