Domains are the three-dimensional building blocks of proteins. An individual domain can be found in a variety of protein architectures that perform unique functions and are subject to different evolutionary selective pressures. We describe an approach to evaluate the variability in amino acid sequences of a single domain across architectural contexts. The ability to distinguish the different evolutionary paths of one protein domain can help determine whether existing knowledge about a specific domain will apply to an uncharacterized protein. Such discrimination can lead to insights and hypotheses about function, as well as guide experimental priorities.We developed and tested our approach on CheW-like domains (PF01584), which mediate protein/protein interactions and are difficult to compare experimentally. CheW-like domains occur in CheW scaffolding proteins, CheA kinases, and CheV proteins that regulate bacterial chemotaxis. We chose 16 protein Architectures that included 94% of all CheW-like domains found in nature. Because some Architectures had more than one CheW-like domain, CheW-like domains were divided into 21 distinct Contexts. The CheW-like domain sequences were closely related within most Contexts; however, one Context was subdivided into three Types. The resulting 23 sequence Types coalesced into five or six Classes of CheW-like domains, which we described in detail.In addition, we created SimpLogo, an innovative method for visualizing amino acid composition across large sets of multiple sequence alignments of arbitrary length. SimpLogo offers substantial advantages over standard sequence logos for comparison and analysis of related protein sequences. The R package for SimpLogo is freely available.