A study was carried out to determine whether bottle-blowing has any positive effects in patients with pneumonia. In a prospective open study 145 adults with untreated community-acquired pneumonia requiring hospitalization were randomized to early mobilization (group A), to sit up and take 20 deep breaths on 10 occasions daily (group B), or to sit up and to blow bubbles in a bottle containing 10 cm water through a plastic tube 20 times on 10 occasions daily (group C). Peak expiratory flow (PEF), vital capacity (VC), forced expiratory volume in 1 sec (FEV1) and serum concentration of C-reactive protein (CRP) were determined on admission, and on days 4 and 42. Fever duration and hospital stay were recorded. In a subset of 16 patients, single breath diffusion capacity of carbon monoxide was measured on 3 occasions. The patients in group A were hospitalized for a mean of 5.3 days, group B for 4.6 days and group C for 3.9 days. Treatment was a significant factor (p = 0.037) in a Cox regression model, with group C significantly better than group A (p = 0.01). The number of days with fever was 2.3, 1.7 and 1.6 in groups A, B and C respectively. These differences were not significant (p = 0.28). No significant differences were found between the groups regarding CRP, PEF, VC, FEV1, or diffusion capacity. Intensive bottle-blowing shortens the hospital stay in patients with pneumonia. The underlying mechanism is not clear.