When the anode potential of a nonoscillating magnetron is lower than the cut‐off potential, the discrete electron‐to‐electron interaction has an effect on the space‐charge distribution which would be negligible if only resulting in fluctuations about a known steady state. However the states proposed by Brillouin, Slater, and others are not steady states in this sense. The electron interaction must, therefore, be expected to produce a steady drift away from such initial states. This paper discusses qualitatively the final steady state to be expected and its dependence on the ratio of anode voltage to cut‐off voltage. Because of the involved nature of the electron distribution function no attempt is made to calculate the space‐charge distribution explicitly.