This study was designed to test the effects of pursed lips breathing (PLB) during exercise in patients with chronic obstructive pulmonary disease (COPD) who did not spontaneously perform PLB. Sixty-nine COPD patients, mean FEV1 (SD) 1.09 (0.5), age 68 (51-83) were recruited to the study. They performed three incremental shuttle walk tests (ISWT). The first walk was designed to identify natural PLBs and the next two walks were performed in a random order; ISWT + PLB or ISWT whilst breathing normally. Measures of respiratory rate (RR), breathlessness and oxygen saturation were taken before and after walks. Data was analysed using the t-test. Fifteen patients demonstrated PLB on baseline ISWT and were therefore excluded from further walking tests although baseline data was included in the analysis. There was no significant difference between walks, mean (SD), 298.5 (173.7) PLB and non-PLB; 292.5 (161.9) nor any difference in dyspnoea. There was a significant reduction in end exercise RR and recovery time with PLB, mean difference (95% CI); 6.2 (4.5-7.9) and 24.9 (2.8-47.0) seconds, respectively. Patients who showed a good response with the PLB walk (41%) had significantly higher baseline breathlessness, Borg score, mean (SD), 1.5 (1.0) versus 0.74 (0.96) (P = 0.02). Natural PLB patients demonstrated lower exercise tolerance on the baseline walk (P = 0.01) and a trend towards greater resting breathlessness than those who did not. This study shows PLB during exercise and recovery results in lower post exercise RR and speeds return to pre exercise breathlessness, compared with exercise and non-PLB. Reductions in RR appear to be greatest in those patients with resting breathlessness.