This study is a case-control study looking to identify factors associated with frequent use of hospital services (emergency care and admissions) in COPD patients. Data from 64 patients with moderate-severe COPD (FEV1/FVC < or = 70, FEV1 < or = 50%) were prospectively collected, 32 cases with high consumption of health resources (COPD-HC) and 32 controls. COPD-HC was defined as a patient diagnosed of COPD requiring during one year: (1) two or more hospitalizations; (2) three or more emergency visits; or (3) one admission and two emergency visits. Patients with COPD and a similar age, FEV1 and PaO2 who required no hospital care during the study year (1998) were randomly selected as controls. Demographic, clinical and socioeconomic data were collected from each subject, and evaluations were made of anxiety, health-related quality of life [measured with the St. George's Respiratory Questionnaire (SGRQ)], nutritional parameters, and different therapeutic aspects. Forced spirometry, resting arterial blood gases, maximal respiratory muscle pressures and a 6-min walking test were measured in all cases. After applying a logistic regression model, the variables that finally proved to be independent predictors of frequent use of hospital services were: treatment with salmeterol, the presence of cardiac arrhythmias, and increased SGRQ scores. The administration of inhaled salmeterol multiplied the risk of having COPD-HC criteria by 27.4 (95%CI: 2.4-308.1), while the presence of arrhythmias multiplied the probability of meeting high consumption criteria by 24.3 (95%CI: 1.7-340.1). For each point of worsened quality of life, the risk of hospital care increased 1.06-fold (95%CI: 1.01-1.10). Although a severity bias related to the presence of long-acting beta2-agonists in the final regression equation cannot be ruled out, the variables associated in our sample to an increased utilization of hospital services are the regular use of inhaled salmeterol, the presence of cardiac arrhythmias, and an impaired health-related quality of life. The use of specific strategies aimed at modulating these aspects could, at least in theory, reduce the number of exacerbations requiring hospital care, with the resultant individual and collective benefits derived.