Noninvasive measurements of somatosensory evoked potentials have both clini
cal and research applications. The electrical artifact which results from t
he stimulus is an interference which can distort the evoked signal, and int
roduce errors in response onset timing estimation, Given that this interfer
ence is synchronous with the evoked signal, it cannot be reduced by the con
ventional technique of ensemble averaging. The technique of adaptive noise
cancelling has potential in this regard however, and has been used effectiv
ely in other similar problems. An adaptive noise cancelling filter which us
es a neural network as the adaptive element is investigated in this applica
tion. The filter is implemented and performance determined in the cancellin
g of artifact for in,vivo measurements on the median nerve. A technique of
segmented neural network training is proposed in which the network is train
ed on that segment of the record Lime window which does not contain the evo
ked signal. The neural network is found to generalize well from this traini
ng to include the segment of the window containing the evoked signal. Both
quantitative and qualitative measures show that significant stimulus artifa
ct reduction is achieved.