River channelization and artificial levees have decreased the hydrologic connectivity of river-floodplain systems around the world. In response, restoration through enhancing connectivity has been advocated to improve the functions of floodplains, but uncertain benefits and the possibility of phosphate release from re-flooded soils has limited implementation. In this study, we measured change in floodplain P, N, and sediment mass balances after restoration along channelized reaches in the lowland Pocomoke River, Maryland USA. Two floodplains (one headwater, one mainstem) restored through partial levee breaches were compared to two additional mainstem floodplains (one natural unchannelized, one unrestored channelized). Potential soluble reactive P (SRP) release from soil cores during experimental laboratory floods; soil P, Fe, and Al fractionation; and deposition and P and N content of sediment were measured before and after the restoration period, as well as in situ inputs and release of SRP and dissolved inorganic N from soils after restorations. Potential SRP release, during both the before and after restoration period, was greatest at the channelized mainstem and restored mainstem sites, lower at the restored headwater site, and small at the natural mainstem site. Both restored sites had smaller potential SRP release after restoration compared to before restoration. In situ SRP release slightly exceeded inputs to soils at connected sites during the post-restoration period, with less net release at the restored sites compared to the natural mainstem site. The magnitude of gross and net SRP release from soils in the field was smaller than, and uncorrelated with, potential SRP release estimated from laboratory experimental floods. Gross soil SRP release rates in the field were predictable using the ratio of soil oxalate-extractable P/Al. Sedimentation inputs of P and N increased at all sites during the post-restoration period, with rates at restored sites intermediate compared to the much higher rates at the natural mainstem site and somewhat lower rates at the channelized mainstem site. These sediment inputs of nutrients were much larger than rates of inorganic P and N release from soils, indicating net trapping of P and N after restoration. Restoring floodplain hydrologic connectivity showed moderate success at increasing the trapping of P, N, and sediment, with relatively little phosphate release, and therefore improving water quality.