Abstract: The qualitative properties of bidirectional associative memory neural networks with delays are studied by introducing a key condition. The key condition connects the bidirectonnal connection weights of the networks, so it could be easy for applications. Under the key condition, it is proved that each of the networks has a unique equilibrium and this equilibrium is globally exponentially stable. If the outer input is periodic, then each of the networks has a unique periodic solution and all other solutions of the network converge exponentially to this periodic solution.