The author previously reported an experimental result of evolutionary reinforcement learning of neural network controllers. In the previous study, a conventional multilayer perceptron was employed in which connection weights were real numbers. In this study, the author experimentally applies an evolutionary algorithm to the reinforcement training of binary neural networks. In both studies, the same task and the same evolutionary algorithm are utilized, i.e. the Acrobot control problem and Evolution Strategy respectively. The differences lie in the memory size per connection weight and the model size of the neural network. The findings from this study are (1) the optimal number of hidden units for the binary MLP was 128 among the choices of 16, 32, 64, 128 and 256; (2) a larger population size contributed better for ES than a greater number of generations; and (3) binary connection weights can achieve comparable control performance while reducing memory size by half.