This paper introduces a dereverberation algorithm based on Linear Prediction (LP) applied to the outputs of an Acoustic Vector Sensor (AVS). The approach applies adaptive beamforming to take advantage of the directional outputs of the AVS array to obtain a more accurate LP spectrum than can be obtained with a single channel or Uniform Linear Array (ULA) with a comparable number of channels. This is then used within a modified version of the Spatiotemporal Averaging Method for Enhancement of Reverberant Speech (SMERSH) algorithm derived for the AVS to enhance the LP residual signal. In a highly reverberant environment, the approach demonstrates a significant improvement compared to a ULA as measured by both the Signal to Reverberant Ratio (SRR) and Speech to Reverberation Modulation Energy Ratio (SRMR) for sources ranging from at 1m to 5m from the array.The effect of reverberation on speech and audio plays an important role in making the sound more natural, but when the level of reverberation increases (beyond = 1s) it significantly degrades the quality of the speech signals in terms of intelligibility due to box effect and distant talker effect [1,2]. Reverberation also leads to altering of the parameters derived for source-filter models of speech typically used in applications such as distant speech recognition and hence degrades their performance [2]. There are several methods that can be used for enhancing speech through dereverberation including beamforming methods; traditional speech enhancement methods; and blind system identification and equalization methods, where the acoustic impulse response of a room is identified blindly and then used to design an equalization filter [2]. These methods can also be categorized into single or multichannel approaches. This paper focuses on the latter methods, which generally result in significant performance gains compared to single channel approaches [2].Estimation of a room impulse response or acoustic transfer function (ATF) is difficult to obtain accurately [2] as it depends on the geometry, the furniture in the room and materials used in the construction of the room. In contrast multichannel algorithms that do not rely on the ATF but instead are based on Linear Predictive Coding have been shown to be highly successful [3][4][5], including the Spatiotemporal Averaging Method for Enhancement of Reverberant Speech (SMERSH) [2] [6] that was designed for a Uniform Linear Array (ULA) of microphones. In this paper, a new multichannel dereverberation method based on adapting the SMERSH algorithm for an Acoustic Vector Sensor (AVS) is proposed. The AVS has three co-located velocity gradient microphones and one omni-direction microphone arranged orthogonally in an area occupying no more than 1cm 3 [7]. The use of gradient sensors allows for precise recording of directional sound component and has been shown to be affective for speech enhancement in additive noise and mildly reverberant environments [8-10] and here it is shown that adapting SMERSH to the AVS provides a superio...