Coastal salt marshes along the northern Gulf of Mexico shoreline received varied types and amounts of weathered oil residues after the 2010 Deepwater Horizon oil spill. At the time, predicting how marsh bacterial communities would respond and/or recover to oiling and other environmental stressors was difficult because baseline information on community composition and dynamics was generally unavailable. Here, we evaluated marsh vegetation, physicochemistry, flooding frequency, hydrocarbon chemistry, and subtidal sediment bacterial communities from 16S rRNA gene surveys at 11 sites in southern Louisiana before the oil spill and resampled the same marshes three to four times over 38 months after the spill. Calculated hydrocarbon biomarker indices indicated that oil replaced native natural organic matter (NOM) originating from Spartina alterniflora and marine phytoplankton in the marshes between May 2010 and September 2010. At all the studied marshes, the major class-and order-level shifts among the phyla Proteobacteria, Firmicutes, Bacteroidetes, and Actinobacteria occurred within these first 4 months, but another community shift occurred at the time of peak oiling in 2011. Two years later, hydrocarbon levels decreased and bacterial communities became more diverse, being dominated by Alphaproteobacteria (Rhizobiales), Chloroflexi (Dehalococcoidia), and Planctomycetes. Compositional changes through time could be explained by NOM source differences, perhaps due to vegetation changes, as well as marsh flooding and salinity excursions linked to freshwater diversions. These findings indicate that persistent hydrocarbon exposure alone did not explain long-term community shifts. IMPORTANCE Significant deterioration of coastal salt marshes in Louisiana has been linked to natural and anthropogenic stressors that can adversely affect how ecosystems function. Although microorganisms carry out and regulate most biogeochemical reactions, the diversity of bacterial communities in coastal marshes is poorly known, with limited investigation of potential changes in bacterial communities in response to various environmental stressors. The Deepwater Horizon oil spill provided an unprecedented opportunity to study the long-term effects of an oil spill on microbial systems in marshes. Compared to previous studies, the significance of our research stems from (i) a broader geographic range of studied marshes, (ii) an extended time frame of data collection that includes prespill conditions, (iii) a more accurate procedure using biomarker indices to understand oiling, and (iv) an examination of other potential stressors linked to in situ environmental changes, aside from oil exposure.