Metagenomic approach, which is the genome analysis on a mixture of uncultured microorganisms, has been recently developed to search for novel and industrially useful genes and to study microbial diversity in a wide variety of environments. We have previously developed a Batch-Learning SOM (BLSOM) for genome informatics, which does not depend on the order of data input. Here we used BLSOM as a novel bioinformatics strategy to unveil and visualize microbial community structures and relative abundance of microorganisms within an environmental sample. The present method is also useful for surveys of microorganisms in extremity environment samples and novel pathogenetic microorganisms in clinical samples.