Background: The 2011 GOLD (Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease [COPD]) consensus report uses symptoms, exacerbation history, and forced expiratory volume (FEV1)% to categorise patients according to disease severity and guide treatment. We aimed to assess both the influence of symptom instrument choice on patient category assignment and prospective exacerbation risk by category. Methods: Patients were recruited from 21 centres in the USA, as part of the COPDGene study. Eligible patients were aged 45-80 years, had smoked for 10 pack-years or more, and had an FEV1/forced vital capacity (FVC) <0·7. Categories were defined with the modified Medical Research Council (mMRC) dyspnoea scale (score 0-1 vs ≥2) and the St George's Respiratory Questionnaire (SGRQ; ≥25 vs <25 as a surrogate for the COPD Assessment Test [CAT] ≥10 vs <10) in addition to COPD exacerbations in the previous year (<2 vs ≥ 2), and lung function (FEV1% predicted ≥50 vs <50). Statistical comparisons were done with k-sample permutation tests. This study cohort is registered with ClinicalTrials.gov, number NCT00608764. Findings: 4484 patients with COPD were included in this analysis. Category assignment using the mMRC scale versus SGRQ were similar but not identical. On the basis of the mMRC scale, 1507 (33·6%) patients were assigned to category A, 919 (20·5%) to category B, 355 (7·9%) to category C, and 1703 (38·0%) to category D; on the basis of the SGRQ, 1317 (29·4%) patients were assigned to category A, 1109 (24·7%) to category B, 221 (4·9%) to category C, and 1837 (41·0%) to category D (κ coefficient for agreement, 0·77). Significant heterogeneity in prospective exacerbation rates (exacerbations/person-years) were seen, especially in the D subcategories, depending on the risk factor that determined category assignment (lung function only [0·89, 95% CI 0·78-1·00]), previous exacerbation history only [1·34, 1·0-1·6], or both [1·86, 1·6-2·1; p<0·0001]). Interpretation: The GOLD classification emphasises the importance of symptoms and exacerbation risk when assessing COPD severity. The choice of symptom measure influences category assignment. The relative number of patients with low symptoms and high risk for exacerbations (category C) is low. Differences in exacerbation rates for patients in the highest risk category D were seen depending on whether risk was based on lung function, exacerbation history, or both. © 2013 Elsevier Ltd.