Background: Late-onset asthma has been shown to be more severe than early-onset asthma in clinic-based studies. However, population-based studies are scarce, and the predictors of severity have been less studied. Objectives: To determine asthma severity and severity predictors regarding age at onset. Methods: A cross-sectional questionnaire survey was conducted among parents of children from 94 schools in Taiwan in 2004. Asthma severity was defined as short-acting β2-agonist (SABA), inhaled corticosteroid (ICS) and health care use in the last year. Information on age at onset, demographics, heredity and home exposure was collected. Ordered logistic or logistic regression was used for determining the associations between risk factors and severity. Results: Participants aged 26-50 years were included, resulting in 21,057 (67.8%) participants. Among them, 449 reported ever having had physician-diagnosed asthma, and 381 of those subjects answered the question on age at asthma onset. The risks of rescue SABA, ICS and health care use were generally higher among late-onset (13-50 years) than early-onset (0-12 years) asthmatics. Use of SABA and health care increased from childhood-onset, adolescent- or young adult-onset to adult-onset asthma. Allergic rhinitis was positively associated with SABA use (OR, 9.08; 95% CI, 1.06-77.99) and ICS use (OR, 5.08; 95% CI, 1.47-17.52) in early-onset asthma. Dehumidifier use was negatively associated with SABA use (OR, 0.50; 95% CI, 0.29-0.87) and ICS use (OR, 0.38; 95% CI, 0.19-0.78) in late-onset asthma. Conclusions: In adults, late-onset asthma was more severe than early-onset asthma. Severity, as indicated by SABA and ICS use, was positively associated with allergic rhinitis in early-onset asthma and negatively associated with dehumidifier use in late-onset asthma.