In France, home oxygen therapy for patients with chronic obstructive pulmonary disease (COPD) is carried out by nonprofit associations (NP) or profit-making health organisations (PM). In a retrospective pragmatic approach we analysed the costs and the effectiveness of these 2 types of structures delivering oxygen at home. Between July 1985 and March 1994, 234 patients were involved in the survival study (chosen as an effectiveness indicator), 24% in PM and 76% in NP. The economic appraisal was performed, from the insurer's point of view, on a representative sample of 61 patients and analysed in detail all the ambulatory costs for respiratory care. Patient survival was similar in both types of structures (Cox model). Oxygen therapy represented the largest share of the total ambulatory cost (81.6% in PM and 72.1% in NP). The NP structures were less costly for reasons linked to their preference for concentrator (p = 0.004 in a Wilcoxon test), all the other direct costs being non-statistically different. NP structures had a significant influence on a low level of ambulatory costs (adjusted OR = 10.98, p = 0.0004) in logistic regression. As oxygen treatment plays an important role in the variation of costs, further pragmatic studies should help to better understand what are the real motivations to choose one mode of oxygen administration more than an other and should determine factors that may sometimes lead physicians not to comply with clinical guidelines (actually a quarter of the patients did not have a PaO2 < 60 mmHg).