An optimized design model of Q-switched lasers is established with neural networks (NN). Obtain the laser outputs with the numerical solution for the rate equations when given cavity parameters and pumping parameters,and these data will form the training set of neural networks. A back-propagation (BP) neural network is established after being trained based on the relationship between the laser structure parameters,pumping parameters and laser outputs. A quick design of Q-switched laser with the network to predict the laser structure parameters and pumping parameters according to required laser outputs can be made while the training time is 2.96 s for computer configuration of Pentium 4 2.4 GHz. Test data and experimental results show that the design standard deviation of peak output power,beam quality and pulsewidth are respectively 3.1%,3.4% and 5.8%. The results give good agreement and the errors are small.