The severity of chronic obstructive pulmonary disease (COPD) is currently assessed using a single physiological measurement, the forced expiratory volume in 1s (FEV1). COPD, however, has complex effects on other aspects of respiratory function, and in many patients is associated with important systemic changes. We hypothesized that a multidimensional staging system for COPD could provide a more complete assessment of the disease's impact. We considered over 40 potential staging variables, evaluating them according to sensitivity to change, measured reproducibly, independence of the information they provide and prognostic value. We finally selected three: FEV1 (including arterial blood gas measurements when FEV1 falls below 35% predicted), Medical Research Council dyspnea scale and body mass index (BMI). Each measure correlates independently with prognosis in COPD, is supported by a significant body of literature and serves as a surrogate for other potentially important variables. We then used principal components analysis (PCA) to determine the degree of association between 30 of the potential variables measured in 813 stable COPD patients. Using PCA, six groups of measurements defined independent categories of patient information: pulmonary function (including FEV1), symptoms of cough and sputum, dyspnea, health status, bronchodilator reversibility and BMI. These include the three principal variables selected for the staging system. Although the staging boundaries were based on existing literature, they have proven useful in predicting survival. We conclude that a multidimensional grading system is useful to assess the impact of COPD.