The highquality network reconstruction from metabolites, enzymes and reactions is the first step for the study on metabolic networks. However, the previous reconstruction approaches have some disadvantages. For example, the data contains redundancy and could hardly be updated in time. Besides, the data retrieval is always timeconsuming. In this paper, we propose a new bottomup approach for the organismspecific metabolic network reconstruction. The web service of KEGG undertakes the data to be correct and uptodate. The data selection mechanism of KEGG/PATHWAY ensures the reconstructed networks reliable. The whole approach is implemented using Java, which can be used on any platform independently. The MySQL database is deployed to map the remote data locally, which greatly shortens the elapsed time for the data retrieval. As is shown in the evaluation, this method can not only ensure the data for the network reconstruction is accurate and uptodate, but also shorten the time in the case that many metabolic networks need reconstructing repeatedly.