<b><i>Background:</i></b> Lung diffusing capacity (D<smlcap>LCO</smlcap>) and lung volume distribution predict exercise performance and are altered in COPD patients. If pulmonary rehabilitation (PR) can modify D<smlcap>LCO</smlcap> parameters is unknown. <b><i>Objectives:</i></b> To investigate changes in D<smlcap>LCO</smlcap> and ventilation inhomogeneity following a PR program and their relation with functional outcomes in patients with COPD. <b><i>Methods:</i></b> This was a prospective, observational, multicentric study. Patients were evaluated before and after a standardized 3-week PR program. Functional assessment included body plethysmography, D<smlcap>LCO</smlcap>, transfer factor (KCO) and alveolar volume (V<smlcap>A</smlcap>), gas exchange, the 6-min walking test (6MWT) and exercise-related dyspnea. Patients were categorized according to the severity of airflow limitation and presence of ventilation inhomogeneity, identified by a V<smlcap>A</smlcap>/TLC <0.8. <b><i>Results:</i></b> Two hundred and fifty patients completed the study. Baseline forced expiratory volume in 1 s (FEV₁) % predicted (mean ± SD) was 50.5 ± 20.1 (76% males); 137 patients had a severe disease. General study population showed improvements in 6MWT (38 ± 55 m; p < 0.01), D<smlcap>LCO</smlcap> (0.12 ± 0.63 mmol × min^-1 kPa^-1; p < 0.01), lung function and dyspnea. Comparable improvements in D<smlcap>LCO</smlcap> were observed regardless of the severity of disease and the presence of ventilation inhomogeneity. While patients with V<smlcap>A</smlcap>/TLC <0.8 improved the D<smlcap>LCO</smlcap> increasing their V<smlcap>A</smlcap> (177 ± 69 ml; p < 0.01), patients with V<smlcap>A</smlcap>/TLC >0.8 improved their KCO (8.1 ± 2.8%; p = 0.019). The latter had also better baseline lung function and higher improvements in 6MWT (14.6 ± 6.7 vs. 9.0 ± 1.8%; p = 0.015). Lower D<smlcap>LCO</smlcap> at baseline was associated with lower improvements in 6MWT, the greatest difference being between subjects with very severe and mild D<smlcap>LCO</smlcap> impairment (2.7 ± 7.4 vs. 14 ± 2%; p = 0.049). <b><i>Conclusions:</i></b> In COPD patients undergoing a PR program, different pathophysiological mechanisms may drive improvements in D<smlcap>LCO</smlcap>, while ventilation inhomogeneity may limit improvements in exercise tolerance.