Abstract The signal strength from LDF probes positioned in perfused muscle can be altered by vasoconstrictors despite total flow being maintained constant. Apart from redistribution of flow via collateral channels outside the region of measurement, the change in LDF signal may arise because the vasoconstrictors have switched flow to vessels of different architecture or altered the architecture of the blood vessels being perfused. Thus we have examined the effect of tube architecture on LDF signal using polymer tubes of 250, 100, and 50 μm internal diameter. At 3% hematocrit the LDF signal was linear for each of the three tube sizes from 10 to 80 μl/h. The signal strength was greatest from the smallest tube and least from the largest tube. For a single tube (100 μm) that doubled back on itself twice to cross the field of measurement three times, the LDF signal at any flow (10–80 μl/h, hematocrit 3%) was approx threefold greater than that for the same tube crossing the field of measurement once. The effect of progressively switching flow (constant at 120 μl/h, hematocrit 9%) from five to one tube in a manifold of five tubes (100 μm) gave rise to a progressive increase in signal. It is concluded that LDF signal derives predominantly from nonvectorial cell speed and less from cell number. Thus any agent that alters the architecture has the potential to alter the LDF signal.