Neural networks have been trained to classify different types of plasma disruptions in a Tokamak experiment using several diagnostic signals as input. Tests refer to data collected from disrupted pulses that occurred during four years of JET experiments. The results show the feasibility of a reliable neural network classifier.