Indicator dilution techniques were applied to measure mean transit time of fluorescently labeled red blood cells (RBCs) (TTRBC) and plasma (TTpl) between functionally paired arterioles and venules (A-V) in cremaster muscle (rat) for normal RBCs and cells hardened by in vitro incubation in graded concentrations of glutaraldehyde. Dispersion of a bolus introduced into the contralateral femoral artery permitted computation of TT by cross-correlation of fluorescence intensity-time curves in A-V pairs. Parallel in vitro assessments of RBC deformability were made by filtration through 5-μm pore Nuclepore filters to express deformability in terms of the ratio of resistance to flow through a pore with RBCs present to that with suspending medium alone, β. The average microvascular hematocrit (Hmicro) normalized with respect to systemic hematocrit (Hsys) was calculated from TTRBC and TTpl. For 26 A-V pairs of the third and fourth orders of branching, TTRBC averaged 0.63 sec for normal control cells (β = 2.61), and TTpl averaged 0.85 sec with an average TTRBC/TTpl equal to 0.85. The corresponding value of Hmicro /Hsys was significantly <1 and averaged 0.87. This greater value of Hmicro /Hsys compared to direct measurements in the literature was attributed to the unique ability of the indicator dilution technique to account for red cell flux throughout the network. For hardened RBCs with β < 10, TTRBC/TTpl and Hmicro /Hsys increased on average 30%, but were weakly correlated with increasing β due to redistribution of RBCs throughout pathways of lesser resistance. However, as β rose from 10 to 20, these pathways became overwhelmed by hardened RBCs and TTRBC/TTpl increased threefold due to retardation of the RBC flux, with a concomitant rise in Hmicro /Hsys. These results clearly demonstrate the extent to which diminished RBC deformability of a magnitude found in clinical disorders may affect microvascular perfusion. © 1993 Academic Press.