We tackle the problem of object completion from point clouds and propose a novel point cloud completion network using a feature matching strategy, termed as ASFM-Net. Specifically, the asymmetrical Siamese auto-encoder neural network is adopted to map the partial and complete input point cloud into a shared latent space, which can capture detailed shape prior. Then we design an iterative refinement unit to generate complete shapes with fine-grained details by integrating prior information. Experiments are conducted on the PCN dataset and the Completion3D benchmark, demonstrating the state-of-the-art performance of the proposed ASFM-Net. The codes and trained models will be open-sourced.