Abstarct: Random noise attenuation plays an important role in the seismic signal processing. Noise in seismic data can be devided into two categories of coherent and incoherent or random noise. Incoherent noise is usually characterized by random oscillation in seismic data over the entire time and frequency. Many algorithms have been developped to address the random noise attenuation problem in seismic records. Different methods are baxsed on different assumptions about the signal. Random noise attenuation in seismic shot gathers is an important processing step which enables other processes that rely on the input data to work more effectively such as deconvolution and pre-stack imaging. This work aims to describe a method which adaptively performs random noise attenuation in the shearlet domain of sliding 2-dimensional windows.Weconducted different denoising experiments on real and synthetic shot gathers with different signal-to-noise ratio. The results show that the proposed method is capable of suppressing the random noise while preserving the signal on the pre-stack gathers. Similar to the curvelet transform, the shearlet transformis multi-dimensional transform that has been developed for multi-scale directional analysis. The shearlet transform has been succefully applied to low-level image processing tasks such as image denoising and image deconvolution as well as high-level processing tasks such as edge detection and feature extraction. The shearlet transform has also shown promising results when applied in seismic data interpolation problems. We propose to apply the shearlet transform for denoising of seismic shot gathers. In particular, we use the transform formulation for fast and efficient computation of the transform coefficient. We present an adaptive strategy for threshold estimation and selection. Finally, we address the problem of edge effects and alleviate reconstruction artifacts by applying the method on sliding 2-dimensional windows of the data.