1. Resting pulmonary blood flow (Q), using the uptake of the soluble inert gas Freon-22 and an indirect estimate of lung tissue volume, has been estimated during breath-holding (Qc) and compared with direct Fick cardiac output (Qf) in 16 patients with various cardiac disorders. 2. The effect of breath-hold time was investigated by comparing Qc estimated using 6 and 10 s of breath-holding in 17 patients. Repeatability was assessed by duplicate measurements of Qc in the patients and in six normal subjects. 3. Qc tended to overestimate Qf, the bias and error being 0.09 l/min and 0.59, respectively. The coefficient of repeatability for Qc in the patients was 0.75 l/min and in the normal subjects was 0.66 l/min. For Qf it was 0.72 l/min. There was no significant difference in Qc measured at the two breath-hold times. 4. The technique is simple to perform, and provides a rapid estimate of Q, monitoring acute and chronic changes in cardiac output in normal subjects and patients with cardiac disease.