Patients hospitalised for exacerbations contribute significantly to the total chronic obstructive pulmonary disease (COPD)-related healthcare costs. This study aimed to determine the resource use and costs of exacerbations by exacerbation-severity and to identify risk factors for hospitalisation. Exacerbations and the details of all associated healthcare utilisation were recorded as part of a prospective cost-effectiveness analysis linked to two randomised controlled trials comparing tiotropium with ipratropium in 519 patients with stable COPD at study entry in the Netherlands and Belgium. Exacerbation-severity was rated by the physician. A Cox proportional hazards analysis was performed to identify independent risk factors of hospitalisation. Covariates that entered this analysis were smoking status, pack-years, body mass index, number of concomitant diseases, number of concomitant medications, use of inhaled steroids, physician visits prior to trial, FEV1% predicted, quality of life, baseline dyspnea index (BDI) and treatment arm. The mean number of exacerbations per patient was 0.70 (95%-CI:0.60, 0.81). About 10% of the exacerbations was severe, 47% moderate and 43% was mild. The mean costs of these exacerbations were Euro 4007 (95%-CI:2004, 6011), Euro 579 (390, 768) and Euro 86 (49, 124), respectively. In addition to treatment arm, a body mass index below 18.5 (RR:3.62), each additional concomitant diagnosis (RR:1.40) and a decrease of 1 point in the baseline dyspnea index (RR:1.18) were significant risk factors of hospitalisation. Exacerbations that were associated with a hospitalisation accounted for 90% of the total costs of exacerbations. Underweight, history of concomitant diseases and increased dyspnea (BDI score) are factors that are likely to identify patients who are at increased risk for generating high costs due to hospitalisation.