Under a complex environment, robust visual tracking is a challenging problem due to the presence of noise, occlusion and variation of illumination. The recently proposed compressed sensing based tracker, has overcome these difficulties to a remarkable extent. Under such framework, it actually assumes that the coding residual follows Gaussian or Laplacian distribution, which may not be general enough to model the coding errors in real application scenarios. Motivated by the recently proposed new scheme for face recognition, namely robust sparse coding (RSC) [6], we propose a weighted sparse coding residue tracker. It is aimed to account for the outliers, beyond the scope of Gaussian and Laplacian noise, which is hard to describe in mathematics but common in practice. The online tracking optimization problem is solved via an efficient iteratively reweighted sparse coding algorithm. Experiments on video sequences show the excellence of the proposed tracker.