Single-breath carbon monoxide diffusing capacity in the whole lung (DLCO) and per unit alveolar volume (DLCO/VA), as expressed in percentage of normal values, gave discordant results when VA of the patients was abnormal. It was hypothesized that normal reference values were inappropriate to interpret data collected in such patients. To substantiate this hypothesis, DLCO and DLCO/VA were measured in four groups: (1) normal volunteers in whom both indices were measured at five different VA; (2) patients with high VA; (3) emphysematous patients; and (4) patients with diffuse interstitial lung diseases (DILD). In normal subjects, DLCO increased and DLCO/VA decreased with VA. In patients with overinflated lungs, the percentage of DLCO was more increased than DLCO/VA. In the emphysematous patients, both indices were equally decreased. In patients with DILD, DLCO was significantly more decreased than DLCO/VA in those suffering from a restrictive pattern. Theoretical values were re-calculated taking into account their true VA and using the relationships observed between DLCO, DLCO/VA and VA. The divergences between DLCO and DLCO/VA were strongly minimized. Therefore, the authors suggest the need to correct theoretical formulas in the presence of a restrictive pattern.