Six indexes for diagnosing uneven ventilation by tracer gas washout were studied. The indexes were lung clearance index, mixing ratio, Becklake index, multiple-breath alveolar mixing inefficiency, moment ratio, and pulmonary clearance delay, all of which increase with impaired pulmonary gas mixing. In model lung tests, indexes that compared the actual washout curve with a calculated ideal curve (mixing ratio, multiple-breath alveolar mixing inefficiency, and pulmonary clearance delay) were unaffected by changes in tidal volume and series dead space, whereas the others varied markedly. In both spontaneously breathing and mechanically ventilated patients all indexes showed a significant difference between smokers and nonsmokers (P less than 0.002), but the indexes were somewhat different in their assessment of different ventilatory patterns. However, the mean value for all indexes, with the exception of mixing ratio, was smallest with a fast insufflation followed by an end-inspiratory pause. Any of the indexes may be useful if its limitations are recognized, but mixing ratio, multiple-breath alveolar mixing inefficiency, and pulmonary clearance delay seem preferable, because they are not affected by changes in tidal volume and dead space fraction.