We examined fish assemblages in tidal salt marsh creeks in Delaware Bay in order to evaluate their response to treatment forPhragmites removal following initial treatment in 1996. In Alloway Crrek, a tributary to Delaware Bay, reference creeks draining marsh of untreatedPhragmites or naturally occurringSpartina were compared with creeks in marshes treated forPhragmites removal. These reference and treated creeks occur in close proximity and share many characteristics including salinity, temperature, dissolved oxygen, and turbidity, although creeks inPhragmites sites differed slightly in bathymetry. We analyzed a time series of otter trawl collections (22 monthly sample periods from 1999 to 2001) for differences in juvenile fish assemblage among creeks with different vegetation history. Periodically, young-of-the-year (YOY) and age 1+ white perch (Morone americana), YOY spot (Leiostomus xanthurus), YOY Atlantic menhaden (Brevoortia tyrannus), and other species were relatively more abundant atPhragmites sites, but other dominant species were preiodically abundant at all sites. Among-treatment differences based on principal response curves analysis accounted for about 19% of the total species variation, but differences varied widely among sample periods and there is little or no indication of a trend over the 3-yr period. Larger collections were often associated with subtidal structure, which was more common atPhragmites sites and potentially represents a sampling artifact. Assemblages of creeks with differing vegetation history differ weakly but recognizably, suggesting slow or little response to treatment, at least based on otter trawl collections in subtidal marsh creeks.