We describe the theory of a technique for the noninvasive measurement of organ blood flow which is based on the principle of fractionation of cardiac output and is applicable with any recirculating gamma emitting tracer. The technique effectively determines the count rate that would be recorded over the organ if the tracer behaved like radiolabelled microspheres and was completely trapped in the organ's vascular bed on first pass. After correction for organ depth, the estimated first pass activity plateau, expressed as a fraction of the injected dose, is equal to the organ's fraction of the cardiac output (CO). By extending the theory, organ extraction fraction of extractable tracers or mean transit time of nonextractable tracers can be measured. Renal blood flow (RBF) to the normal left kidney in 18 subjects without evidence of renal disease was estimated by the technique to be 10.4% CO (S.D. 1.2), for the right kidney 9.0% CO(S.D. 1.1), and for both, 19.4% CO (S.D. 1.5). In a separate series of eight patients, RBF values were highly reproducible when obtained on successive days (standard deviation of change: 0.67% CO for the left kidney, 0.63% CO for the right and 0.78% CO for both). The extraction fraction of DTPA (filtration fraction) was 8.5% (1.7) in the left kidney and 9.7% (2.1) in the right kidney. This parameter was less reproducible than RBF, with standard deviations of the changes equal to 1.17% for the left kidney and 1.31% for the right. Stable, well-functioning, renal allografts in children had a mean blood flow of 20.6% CO (S.D. 3.7). Transplant blood flow in eight patients with rejection was less than 5.2% CO. In patients without splenomegaly, splenic blood flow, splenic extraction fraction of sulphur colloid and splenic red cell mean transit time were 4.3% CO (S.D. 0.9), 41% (S.D. 13.8), and 36.9 s (S.D. 4.6), respectively.