Stable isotope labeling of microbial taxa of interest and their sorting provide an efficient and direct way to answer the question “who does what?” in complex microbial communities when coupled with fluorescence in situ hybridization or downstream ‘omics’ analyses. We have developed a platform for automated Raman-based sorting in which optical tweezers and microfluidics are used to sort individual cells of interest from microbial communities on the basis of their Raman spectra. This sorting of cells and their downstream DNA analysis, such as by mini-metagenomics or single-cell genomics, or cultivation permits a direct link to be made between the metabolic roles and the genomes of microbial cells within complex microbial communities, as well as targeted isolation of novel microbes with a specific physiology of interest. We describe a protocol from sample preparation through Raman-activated live cell sorting. Subsequent cultivation of sorted cells is described, whereas downstream DNA analysis involves well-established approaches with abundant methods available in the literature. Compared with manual sorting, this technique provides a substantially higher throughput (up to 500 cells per h). Furthermore, the platform has very high sorting accuracy (98.3 +/- 1.7%) and is fully automated, thus avoiding user biases that might accompany manual sorting. We anticipate that this protocol will empower in particular environmental and host-associated microbiome research with a versatile tool to elucidate the metabolic contributions of microbial taxa within their complex communities. After a 1-d preparation of cells, sorting takes on the order of 4 h, depending on the number of cells required.