The use of dynamic adaptation gain/learning rate (DAG) for improving the performance of gradient type adaptation/learning algorithms will be discussed. The DAG is an ARMA (poles-zeros) filter embedded in the gradient type adaptation/learning algorithms and generalizes the various improved gradient algorithms proposed in the literature. After presenting the DAG algorithm and its relation with other algorithms, its design will be summarized. Strictly Positive Real (SPR) and PR conditions play an important role both for the design of the DAG as well as for the stability of the adaptive/learning systems using a DAG.The potential of the DAG is illustrated by experimental results obtained on a relevant adaptive active noise control system (ANC).