We considered whether a slow single breath with neither breath-holding nor carefully controlled flows could provide estimates of lung transfer factor for CO (TLCO) similar to those obtained with the usual standardized single breath technique. This technique requires actual flow rates and volume variations to be taken into account [10], as well as the use of a fast CO analyser and computerized calculations. TLCO values found with this method (TLCOsb) for 5 normal subjects and 29 patients with various respiratory diseases did not differ from those obtained with the standardized test (p less than 0.001). TLCO was also measured during exhalation only, by the use of a single compartment, constant TLCO equation and a computational procedure which provided a mean TLCO value for a given expired volume range (TLCOex). A unique TLCOex was sufficient to account for the whole exhalation in normal subjects and certain patients. In most patients two TLCOex were necessary, one for large lung volume after dead space washout and the other one accounting for the second half of expiration until closing volume. Most TLCOex were larger than TLCOsb calculated during the same slow breath. This over-estimation was found to be correlated (p less than 0.001) with the phase III argon slope. In patients where two TLCOex values were required to describe the exhaled CO course, we found that TLCOex decreased with lung volume. This decrease was also correlated with the argon slope (p less than 0.001). The observed difference between TLCOsb and TLCOex values and the decrease of TLCOex with lung volume probably reflect inhomogeneous ventilation distribution.