SUMMARY1. Normal dogs were trained to adopt a laterally symmetric stance on a horizontal platform. Sinusoidal oscillation of the platform in the cephalocaudal direction caused the dogs to adopt a strategy of response which would keep them from falling down during the period of imposed motion or perturbation.2. A Fourier analysis of the response variables provided a quantitative measure of the distortion in the induced movement at the various hind leg joints and in the motion of the body. Certain aspects of the distortion could be accounted for by recognized random events such as drift and panting. The remainder of the distortion was task related and therefore provided evidence that the dog's postural control system behaved essentially as a non-linear system.3. The motion of the body was less distorted than the motion at the joints. The frequency response of the body motion resembled that of a second order linear system, but the amplitude of the body motion did not vary in constant proportion with changes in the amplitude of the input. Further, the phase relation between body motion and platform motion was not constant when the amplitude of the input was changed. Therefore, it was concluded that the control of the body position was probably non-linearly related to the input disturbance but that the low pass filter characteristics tended to minimize the appearance of distortion in that system output.4. Control of the position of the hind limb is related to the control of the torque generated at the hind limb joints. To the extent that joint angle and change in joint angle are related to the torque at a joint, the distortion of the motion at the joints clearly demonstrates that control of the hind limb during this postural task was non-linearly related to the sinusoidal input.II-2 J. M. BROOKHART AND R. E. TALBOTT 5. The uniformity of the response parameters, as assessed from the Fourier coefficients, indicated that all of the tested dogs adopted the same or nearly the same strategy for solving the problem of adjusting their postural control in response to the perturbation. Therefore, a reasonable hypothesis for future testing is that the central programme which generates this particular postural response will be structured similarly from one dog to the next.