Biofilm formation and removal from dead-ends is a particularly difficult and understudied area of water distribution system biology. In this work, we have built a model drinking water distribution system to probe the effect of different hydrodynamic flow regimes on biofilm formation and removal in the main pipe and in the dead-end. The test rig was built to include all major drinking water distribution system components with materials and dimensions used in standard plumbing systems. We have simulated the effect of stagnant, laminar, turbulent, and intense turbulent flushing conditions on the growth and removal of biofilms from the main pipe and the dead-end. The growth of the biofilm in the main pipe was not prevented at a volumetric flow rate of 9.4 L min−1 and flow velocity of 2 m s−1. Mature biofilms were more difficult to remove. Biofilms grown under shear stress conditions could withstand significantly higher shear stresses than those to which they were exposed to during growth. The biofilms grew twice as fast in the dead-end when flow in the main pipe was turbulent compared to stagnant conditions. Biofilms in the dead-end were not affected by the flushing conditions in the main pipe (Q = 52 L min−1, Re = 9.0 · 104). The computational fluid dynamics simulation suggests that biofilms cannot be hydrodynamically removed from the dead-end at depths that are larger than one pipe diameter. Biofilms beyond this limit present a possible source for reinoculation and recolonization of the rest of the water distribution system.