In this paper, an ensemble of neural networks based incremental learning algorithm with weights updated voting is described. The algorithm defines the class kernel function of the training database of the component neural network in the ensemble. The voting weights are updated based on the distance between the test instance and the kernel function. This method can adaptively update the voting weights according to the classification performance of the component neural network on the test pattern and it is more optimal than the stable weights voting strategy. Experimental results show that the ensemble of neural networks based incremental learning algorithm with weights updated voting is more promising than that with stable weights voting rule.