Reversible watermarking for digital images has become an interesting research area as the distortion introduced by watermarking is completely removed after extracting the hidden data from the cover image. One of the most important technique used for reversible image watermarking is Prediction Error Expansion technique (PEE). In this paper, a comparison between different prediction-based reversible watermarking techniques is conducted to evaluate their performance in terms of embedding capacity and imperceptibility. In addition, the solution of overflow and underflow provided in each technique is studied