Single-particle electron microscopy (EM) is a powerful tool for studying the structures of large biological molecules. However, the achievable resolution does not always allow for direct recognition of individual protein domains. Labels that can be visualized by EM have been developed for protein termini, but tagging internal domains remains a challenge. We describe a robust strategy for determining the position of internal sites within EM maps, termed domain localization by RCT sampling (DOLORS). DOLORS uses monovalent streptavidin added posttranslationally to tagged sites in the target protein. Internal labels generally display less conformational flexibility than terminal labels, providing more precise positional information. Automated methods are used to rapidly generate assemblies of unique 3D models allowing the attachment sites of labeled domains to be accurately identified and thus provide an overall architectural map of the molecule.