Ventilation-perfusion (Va/Q) inequality was measured by a multiple inert gas elimination method in 6 asymptomatic patients with asthma and in a seventh patient during a severe asthmatic episode. Measurements were made before and at 5-min intervals after administration of aerosolized isoproterenol. All patients had some residual airway obstruction as measured during forced expirations. All except one patient had clearly bimodal distributions of Va/Q ratios during all phases of the study, as confirmed by an extensive exploration of distributions compatible with each set of inert gas data. One mode lay within the normal range of Va/Q, but the other, containing 19.8 per cent of the cardiac output on the average, was centered on a Va/Q ratio of only 0.07. There was essentially no shunt. Five min after the administration of isoproterenol, the blood flow to the low Va/Q mode approximately doubled, accounting for the observed decrease in arterial PO2. Breathing 100 per cent O2 had little effect on the distribution. The presence of a bimodal distribution of Va/Q ratios without shunt suggests that collateral ventilation may be an important mechanism determining the distribution of Va/Q ratios and preventing the development of shunts. This study also showed that in some asymptomatic asthmatic patients, as many as one half of the lung units may lie behind completely closed airways and have very low but finite Va/Q ratios as a result of collateral ventilation.