Subduction zones are the regions on Earth capable of generating the largest earthquakes. Our current earthquake catalogs are limited in time, and only a handful of magnitude 9 events have been recorded or documented. Therefore, it is crucial to develop methods to infer which subduction zones are capable of producing very large magnitude earthquakes. For this purpose, subduction zones are typically broken into segments (sources) during the modeling process. This allows modelers to assign different rates and maximum magnitudes to certain sections of the subduction zone. The Aleutian megathrust is one example of a subduction zone that is modeled using this approach. In previous U.S. Geological Survey (USGS) hazard maps, the Alaskan megathrust was segmented using coupled or locked zones, geologic data, and historic earthquakes (Wesson et al., 1999, 2007). In this study, we adapt a Bayesian approach developed to examine seismotectonic zoning to determine the location of segment boundaries (Le Goff and Fitzenz, 2012) to the particular geometry of the Alaskan megathrust. This approach examines a predefined geographic area for contrasts in seismic activity rates, and finds the marginal probability density function for the location of the limit between zones. The advantages of this approach are (1) that it is reproducible and results can be expanded to include additional earthquake catalog data over time, and (2) other data types can be used to build prior distributions for the location of segment boundaries. Using this approach, we find evidence for two robust segment boundaries along the Aleutian megathrust; these are in different locations than the segment boundaries used in the 2007 USGS hazards maps and are supported by recent interpretations of historical events.