There has been much interest in recent years on neural network based control of non-linear dynamic processes and also on their use in multiple model adaptive control schemes. This paper reports on performance enhancements to one class of such control laws, including their extension to multiple model adaptive control.