There is a need for a reliable counting method as an alternative to a standard weir to assess steelhead trout (Oncorhynchus mykiss) escapement in remote streams in Southeast Alaska. The Alaska Department of Fish and Game tested a Dual Frequency Identification Sonar (DIDSON) in Peterson Creek to count a small stock (N approximately 200) of ocean-maturing steelhead in 2009. The DIDSON was operated in an 8-m wide section of stream, and provided near video quality images of steelhead. Detection of steelhead by the DIDSON allowed for enumeration, direction and time of passage, length estimation, and the cross-sectional range of travel from each image. Analysis of the images collected continuously during the field season resulted in detection of 747 steelhead that swam upstream past the DIDSON, and 445 that swam downstream. Daily upstream/downstream behavior (milling/searching) appears to have artificially inflated the count of adult steelhead at the DIDSON site. To account for this behavior, we used a Decision Support Tool developed by the National Marine Fisheries Service Southwest Fisheries Science Center that matched images of adult steelhead moving in opposite directions using elapsed time, length, and group size to determine which steelhead images were likely the same fish. The tool did not work for our purposes. The two major problems were milling and unequal detection of upstream and downstream moving steelhead. The milling behavior of steelhead in Peterson Creek interfered with all aspects of this project. While the DIDSON shows promise for counting iteroparous adult steelhead in small streams in Southeast Alaska, more rigorous protocols must be developed to account for daily migratory behavior of adult steelhead, as well as validation methods.