Fluorescence in situ hybridization (FISH) is a powerful tool for visualizing the spatial organization of microbial communities. However, traditional FISH has several limitations, including limited phylogenetic resolution, difficulty visualizing certain lineages, and the design and optimization of new probes is time consuming and does not scale to the known diversity of microbial life. Here, we present GenomeFISH, a high-throughput, genome-based FISH approach that can differentiate strains within complex communities. Fluorescent probes are generated from the genomes of single cells, which are obtained from environmental or clinical samples through fluorescence activated single-cell sorting. GenomeFISH can distinguish between strains with up to 99% average nucleotide identity and was successfully applied to visualize strains in mock communities and human fecal samples. Given the superior sensitivity and specificity of GenomeFISH, we envisage it will become widely used for the visualization of complex microbial systems.