The distortion of longitudinal pulses in slightly curved elastic rods is discussed in terms of an elementary theory in which the effects of rotary inertia and radial shear are neglected. Moreover, when the wavelength of the pulse is long enough, practically all the bending effects can be justifiably neglected. The experimentally observed strain—time curve produced at one end of a curved brass‐alloy rod by the impact of a pellet shot from an air gun is compared with that obtained theoretically using Fourier synthesis. Good agreement is found between theory and experiment. The main shape of the pulse travelling down the rod is retained even after the rod has turned 90° with respect to the initial direction.