Using three conventional methods and a new method we measured the single-breath diffusing capacity for carbon monoxide [DLCO(SB)] in a group of normal subjects. Whereas the conventional methods calculated DLCO(SB) from a single equation valid only for breath holding, the new method used three equations, one for each phase of the single-breath maneuver, i.e., inhalation, breath holding, and exhalation. We found that while the conventional methods of calculating DLCO(SB) were greatly affected by variations in the way in which the single-breath maneuver was performed and/or the way in which the alveolar gas sample was collected, these variations had little effect on the calculations of DLCO(SB) using the new method. These results were in close agreement with results from a computerized mathematical lung model in which the diffusing capacity did not change with lung volume. We concluded that the new method significantly improves the accuracy and precision of DLCO(SB) measurements while reducing the effects of maneuver variability. For these reasons comparisons of DLCO(SB) values between patients and normal subjects or between two groups with different pulmonary function may be more valid using the new method than using conventional methods.