The operation of a group of parallel-connected, nominally identical switching transistors is prone to difficulties owing to spread in switching times. The fastest and slowest transistors in the group are liable to fail through excessive power dissipation. A method of regulating transistor operation is described in which the application and removal of base current is phase controlled to compensate for differences in switching speeds. The transistors therefore appear identical, in so far as the emitter currents within a group of devices rise and fall in synchronism during switching transitions. This approach offers certain advantages over more conventional solutions to the problem of parallel operation, by allowing higher switching frequencies to be used and by allowing the transistors to pass nearly full-rated current.