A new test generation algorithm based on neural network and artificial bee colony optimization for nonrobust path delay faults is proposed in this paper because the traditional test generation algorithm’s efficiency is low. The algorithm switches digital circuit into equivalent partial leafdag circuit firstly and then constructs the constraint circuit for the equivalent circuit. Hopfield neural network model is constructed for the constraint circuit and energy function is obtained. The test vector for the single stuckat fault of equivalent circuit can be obtained by using artificial bee colony optimization algorithm to solve the minimum of energy function, then the test vector is changed to the test vector pairs for the nonrobust path delay fault of original digital circuit. The experiment results demonstrate that the algorithm’s fault coverage algorithm can achieve 98%, average test generation time is less than 0.8 seconds. 