The stability of the equilibrium point of bidirectional associative memory (BAM) neural networks with delays is studied by constructing a suitable Lyapunov functional and combining with some inequality analysis techniques- On the assumption that the activation functions of neurons are less restrictive than those in t he literature (which may not satisfy Lipschitz condition), a new sufficient cond ition ensuring the global asymptotic stability of BAM neural networks with delay s is derived- The results presented here can be applied to the design of a wider class of neural networks including non-Lipschitz activation functions of neuron s-