In spite of special advantages of IIR filters in active noise and vibration control (ANVC) applications, the multimodal error surface and instability problem of adaptive IIR filters has prevented its extensive use. To alleviate these problems, in this paper, a new RLS-based fast array adaptive IIR filters in ANVC applications is proposed. The algorithm is developed with slow adaptation assumption and by transforming the active noise and vibration control problem to an output-error identification problem. By derivation of the fast-array equivalent form both computational complexity and numerical stability of the proposed algorithm are improved. The geometrical illustration of the algorithm, in a simple case, is also given to unify and complete its mathematical formulation. In spite of low computational complexity of the order O(n), simulation results confirm high convergence speed of the proposed algorithm and also its ability to reach the lower minimum mean square error in comparison with commonly used adaptive IIR algorithms in ANVC systems.