In this paper we present the modelling, design, and experimental testing of a nerve cuff multielectrode system for selective activation of fibres in superficial peripheral nerve trunk regions which is capable of activating fibres in physiological order. The multielectrode system consists of 45 platinum electrodes embedded within a self-curling spiral silicone sheet organized in fifteen longitudinal groups consisting of three electrodes spaced equidistally around the circumference of the cuff. Electrodes in the centre band acted as stimulating cathodes while the two electrodes of the same group in the two outer bands were connected together and corresponded to the position of a particular cathode, serving as anodes to block the nascent action potentials by membrane hyperpolarization. The interpolar distance was 6 mm on both sides, resulting in a total cuff length of about 20 mm. The cuff was constructed with a diameter to fit the size of the dog sciatic nerve. Preliminary animal testing of the nerve cuff was performed on the sciatic nerve of a Bigley female dog. In the 45-electrode stimulation system, biphasic cathodic first pulses with quasitrapezoidal-shaped cathodic and square anodic parts were delivered through the particular group of tripolar electrodes to effect both selective stimulation of motor axons within the gastrocnemius muscle fascicle, and differential block by membrane hyperpolarization. The test was repeated using rectangular cathodic first biphasic current pulses delivered monopolarly on the central electrode of the same group while connected anodes were replaced by a common anode situated elsewhere in the surrounding tissue. In both experiments an isometric torque in the ankle joint elicited by the gastrocnemius muscle was measured and compared. It was shown that tripolar activation with quasitrapezoidal stimulation pulses elicited an isometric torque with a peak value of 0.83 Nm in 65 ms after onset of delivering stimulating pulses in comparison to the monopolar activation with rectangular biphasic pulses where the peak of the same value was observed in 45 ms after onset. Thus, the multipolar cuff stimulating monopolarly provided an effective means of activating motor axons selectively within the gastrocnemius muscle fascicle, while more physiological recruitment of the muscle fibres was evident when stimulating tripolarly.