Biometrics is a method using the physiological or behavioral features of a person for an automated detection and verification of their identity. Biometric-based identification is preferred over traditional methods because a biometric cannot be forgotten, lost or stolen. Nowadays biometrics is becoming a principal method of verification and identification in a networked society. This paper studies the performance of the combination neural network and genetic algorithm for written signature verification in comparison with backpropagation algorithm. Experiments show that using genetic algorithm for neural network parameter learning is better than backpropagation with forgery data available.