A genetic analysis of samples from the chum salmon (Oncorhynchus keta) bycatch from the 2012 Bering Sea walleye pollock (Gadus chalcogrammus) trawl fishery was undertaken to determine the overall stock composition of the sample set. A genetic analysis of chum salmon collected during a test of a salmon excluder device was also conducted. Samples were genotyped for 11 microsatellite markers and results were estimated using the current chum salmon microsatellite baseline. In 2012, genetic samples were collected systematically as part of a special project that commenced in 2011 to reduce sample biases that exist in collections from previous years and have the potential to affect stock composition analysis results. One genetic sample was collected for every 31.5 chum salmon caught in the 98% of the midwater trawl fishery that was sampled. Evaluation of sampling based on time, ___location, and vessel indicated that the genetic samples were representative of the total bycatch. Based on the analysis of 673 chum salmon bycatch samples collected throughout the 2012 Bering Sea trawl fishery, the North Asian stocks dominated the sample set (39%), with moderate contributions from East Asian (20%), Eastern Gulf of Alaska (GOA)/Pacific Northwest (PNW) (18%), and Western Alaska (14%) stocks, and smaller contributions from Upper/Middle Yukon River (7%) and Southwest Alaska (2%) stocks. The estimates for the 2012 chum salmon bycatch sample set differed from the mean of the 20052011 estimates for the two Asian regions, but not for the North American regions. The pattern of changes of regional stock contributions over three time periods in 2012 differed from previous years for some regions. There were some spatial differences in stock distribution; e.g., the East Asian stock contribution was higher in the central Bering Sea than in the southeastern Bering Sea. As with the bycatch samples, the salmon excluder device test samples included fish from all geographic regions despite being collected at small spatial and temporal scales.