We studied the pattern of intermixing of the hepatic arterial and portal venous flows in a perfused rat liver preparation under constant flow (12 ml/min) with intravital epifluorescent microscopy; changes in the steady state extraction ratio of carbon 14-labeled phenacetin and tritiated acetaminophen, probes metabolized primarily in perivenous and periportal regions of the rat liver, respectively; and the spaces accessed by noneliminated reference indicators introduced as a bolus into the hepatic artery and portal vein at different hepatic arterial/portal venous flow regimens of 0:12, 2:10 and 4:8. The sinusoidal velocities for the hepatic arterial- and portal venous (hepatic arterial/portal venous flow at 4:8)-infused fluorescein isothiocyanate-erythrocytes (100 microliters/min) were 327 +/- 78 and 301 +/- 63 microns/sec, respectively, and the velocity for the solely portal venous-perfused liver (12 ml/min) was 347 +/- 74 microns/sec; the flow-weighted sinusoidal velocity was highly correlated to the sinusoidal volume for the dually perfused rat liver. Small but significant decreases in the extraction ratio of [14C]phenacetin (from 0.989 to 0.984 and 0.980) and tritiated acetaminophen (from 0.631 to 0.607 to 0.563), delivered simultaneously into the hepatic artery and portal vein, were observed with an increment of hepatic arterial flow within the same liver preparation; oxygen consumption rate also fell slightly, in parallel fashion. When a multiple-indicator dilution dose containing chromium 51-labeled RBCs, iodine 125-labeled albumin and tritiated water or [14C]urea was injected into the hepatic artery (which accesses both the peribiliary capillary plexus [nonsinusoidal] and the sinusoidal bed) and portal vein (which enters only the sinusoids) at 10-min intervals within each steady state, the blood volume, total albumin space, albumin Disse space, total water and parenchymal cellular water spaces were unchanged after portal venous injection for all hepatic arterial/portal venous flow ratios, suggesting that the arterial flow is ineffective in perturbing average sinusoidal flow dynamics. However, slightly larger total water spaces were obtained with hepatic arterial injection. This excess water space was almost completely accounted for by the "nonsinusoidal" extravascular space associated with the peribiliary capillary plexus; it averaged 0.03 ml/gm and was independent of flow. The anomaly, a reduced flow-weighted sinusoidal velocity for the dually perfused liver, an unchanged diameter of the terminal hepatic venule (32 microns) among the hepatic arterial/portal venous flow ratios and the reduction in the extraction ratio of the drug probes and oxygen consumption rates suggest that some of the arterial flow must have entered the sinusoids somewhat downstream.