Rapid increase of the data size in metagenome researches has raised the demand for new tools to process large datasets efficiently. To accelerate the metagenome profiling process in the scenario of big data, we developed SOAPMetaS, a marker gene-based multiple-sample metagenome profiling tool built on Apache Spark. SOAPMetaS demonstrates high performance and scalability to process large datasets. It can process 80 samples of FASTQ data, summing up to 416 GiB, in around half an hour; and the accuracy of species profiling results of SOAPMetaS is similar to that of MetaPhlAn2. SOAPMetaS can deal with a large volume of metagenome data more efficiently than common-used single-machine tools.