Rationale: Reducing asthma exacerbation frequency is an important criterion for approval of asthma therapies, but the clinical features of exacerbation-prone asthma (EPA) remain incompletely defined.Objectives: To describe the clinical, physiologic, inflammatory, and comorbidity factors associated with EPA.Methods: Baseline data from the NHLBI Severe Asthma Research Program (SARP)-3 were analyzed. An exacerbation was defined as a burst of systemic corticosteroids lasting 3 days or more. Patients were classified by their number of exacerbations in the past year: none, few (one to two), or exacerbation prone (>3). Replication of a multivariable model was performed with data from the SARP-1 1 2 cohort.Measurements and Main Results: Of 709 subjects in the SARP-3 cohort, 294 (41%) had no exacerbations and 173 (24%) were exacerbation prone in the prior year. Several factors normally associated with severity (asthma duration, age, sex, race, and socioeconomic status) did not associate with exacerbation frequency in SARP-3; bronchodilator responsiveness also discriminated exacerbation proneness from asthma severity. In the SARP-3 multivariable model, blood eosinophils, body mass index, and bronchodilator responsiveness were positively associated with exacerbation frequency (rate ratios [95% confidence interval],