Possibly, by using a FEMA flood map and/or soil map (hydrologic soil group or drainage class), approximate locations for floodplains can be determined. In order to use the soil map, you have to get the data from NRCS or statsgo soils and join tables for chorizon. Chorizon will have the hydrologic soil groups and drainage classes. These are very similar so it is up to you to decide. These are a vector so you need to convert the vector into a raster. Once the soil raster is created, you use the combine tool in arcToolbox to combine the soil raster and the DEM. Using the attribute table, you have to see which values indicate the areas of interest corresponding to your floodplain. Hydrologic soil groups C and D indicate soils that are poorly drained to moderately drained. These classes respond to areas on floodplains and creek terraces.
What inputs do you have? If all you have is a terrain DEM then your best bet is to do what the other guy said and look at the soils or look at the effective FIS map.
If you have the terrain data and water surface elevations(either BFE's or model cross sections) there are a number of ways to approach it. The easiest would be to use the TIN Difference tool in the 3dAnalyst extension.
1) convert your terrain DEM to a TIN using 3dAnalyst(3dAnalyst Toolbar->convert->Raster to TIN) Keep the Z tolerance as small as you can(shoot for under 0.5 if working with Feet)
2) convert BFE's or cross sections with associated water surface elevations to a TIN. (3dAnalyst toolbar->Create/Modify TIN) Just check the box next to your data and leave the other settings alone)
3) Tin Difference tool: 3dAnalyst Toolbox->Tin Surface->TIN Difference (TIN 1=WS elevations, Tin 2=terrain)
This will output a polygon shapefile that defines the floodplain. The initial output will be a shape with "0, 1, -1) Delete all polygons with the -1 (its negative because its outside the area defined by the floodplain)