In 69 studies, a definition of an exacerbation was provided (see Table E1 in the online data supplement). Table 1 summarizes the various items used to define an exacerbation. Only one of the definitions included lung function as variable, which points to its limited usefulness in defining exacerbations: FEV1 changes relatively little with COPD exacerbations. A “change in symptoms” was used in 54 studies, but in 14 of these studies no description of the specific symptoms was provided. In 26 studies, the definition of an exacerbation was based completely on this change in symptoms. More recent studies based the definition less frequently on symptoms alone (from 1987 to 1999, 68.5% [11 of 16 studies] vs from 2003 to 2007, 20% [7 of 32 studies]). From 2000 onward, the combination of “a change in symptoms” and “a short course of oral corticosteroids and/or antibiotics” was encountered more frequently (n = 16). In definitions (partly) based on symptoms, the number and type of symptoms varied considerably. Symptoms most frequently used were dyspnea, sputum production, cough, and sputum purulence. Changes in the use of medication (bronchodilators and courses of oral corticosteroids or antibiotics) were incorporated in 29 definitions. Three studies, all published after 2002, defined an exacerbation solely by prescription of a course of oral corticosteroids and/or antibiotics. Whereas the variable “a clinical diagnosis of an exacerbation” was mostly incorporated in studies using exacerbation as an inclusion criteria (7 of 9 studies), the variable “unscheduled medical attention” was mainly used when exacerbations were used as an outcome measure (9 of 10 studies). No other clear distinctions were found between definitions applied in studies in which “exacerbation” was used as either an outcome measure or inclusion criterion. A time period (eg, 24 h) during which the change in symptoms had to be persistent was provided in 22 studies. More recent studies applied this “time period” as frequently as older ones. Finally, a classification in severity of exacerbations based on health-care utilization (mild, moderate, or severe) was used in 13 studies; 11 of these studies were published after 2000. The description of mild and moderate exacerbations differed markedly between studies, but severe exacerbations included hospitalizations in almost all the studies.