Studies of seed bank development have rarely been included in evaluations of wetland restoration. We compared the seed bank of a recently restored tidal freshwater marsh in Washington, D.C., Kingman Marsh, with seed banks of another restored site (Kenilworth Marsh) and two reference marshes (Dueling Creek and Patuxent Marsh). The density and richness of emerging seedlings from Kingman Marsh seed bank samples increased from less than 4 seedlings and 2 taxa/90-cm2 sample in 2000 (the year of restoration) to more than 130 seedlings and 10 taxa/90-cm2 sample in 2003. The most important seed bank taxa at Kingman Marsh included Cyperus spp., Juncus spp., Lindernia dubia, Ludwigia palustris, and the non-native Lythrum salicaria. These taxa are not abundant in most mid-Atlantic tidal freshwater marshes but are almost identical to those described for a created tidal freshwater wetland in New Jersey. Seed banks of both the restored sites contained few seeds of several important species found at the reference sites. Flooding had a significant negative effect on emerging seedling density and taxa density, suggesting that slight decreases in soil elevation in restored wetlands will dramatically decrease recruitment from the seed bank. Because seed banks integrate processes affecting growth and reproduction of standing vegetation, we suggest that seed banks are a useful metric of wetland restoration success and urge that seed bank studies be incorporated into monitoring programs for restored wetlands.