The rapid and reproducible preparation of biological samples is an essential requirement for LC-MS-based clinical metabolomic profiling. Herein, we proposed a novel, automated sample preparation method for serum metabolite LC-MS analysis. This method included a one-step double-filtration process with two filters (30k and 10k MWCO). High sample throughput (up to 96 samples) was achieved in an automated manner by incorporating the process in a liquid handling robotic system. The effectiveness of the developed method was evaluated through comparison with a conventional manual centrifugation-based method using seven human serum samples. A higher metabolite recovery and better reproducibility were achieved in quantitative analysis using the automated double-filtration method. The results suggested that the multiplexed serum sample preparation system can be applied to large-scale untargeted metabolomics studies.