The paper presents the comparative analysis of signature recognition and verification using Neural Networks. Three well-known and widely used Neural Networks viz. Support Vector Machines, Multilayer Perceptron and Radial Basis Function Network have been used. Each network differs from the other in the manner it approaches the signature given for recognition and verification. A signature database is collected using intrapersonal variations for evaluation. For every 6 training examples, 4 are used to test the signatures based on various features like false rejection rate, false acceptance rate, equal error rate, and average error rate. The merits and demerits of all the approaches are evaluated and hence the results of numerical experiments are given and analyzed in the paper. In this paper an off-line Recognition and Verification is done with the objective of performance comparison. The comparison of the three networks is done with respect to the complexity of the structure as well as the accuracy of expected results so that the forgeries can be minimized. Keywords: Neural Networks, offline Signature Verification, Multilayer Perception, Support Vector Machine, Radial Basis Function Network.