Restored riparian wetlands in the Upper Mississippi River basin have potential to remove sediment and nutrients from tributaries before they flow into the Mississippi River. For 3 yr we calculated retention efficiencies of a marsh complex, which consisted of a restored marsh and an adjacent natural marsh that were connected to Halfway Creek, a small tributary of the Mississippi. We measured sediment, N, and P removal through a mass balance budget approach, N removal through denitrification, and N and P removal through mechanical soil excavation. The marsh complex had average retention rates of approximately 30 Mg sediment ha⁻¹ yr⁻¹, 26 kg total N ha⁻¹ yr⁻¹, and 20 kg total P ha⁻¹ yr⁻¹. Water flowed into the restored marsh only during high-discharge events. Although the majority of retention occurred in the natural marsh, portions of the natural marsh were hydrologically disconnected at low discharge due to historical over-bank sedimentation. The natural marsh removed >60% of sediment, >10% of P, and >5% of N loads (except the first year, when it was a N source). The marsh complex was a source of NH₄⁺ and soluble reactive P. The average denitrification rate for the marsh complex was 2.88 mg N m⁻² h⁻¹. Soil excavation removed 3600 Mg of sediment, 5.6 Mg of N, and 2.7 Mg of P from the restored marsh. The marsh complex was effective in removing sediment and nutrients from storm flows; however, retention could be increased if more water was diverted into both restored and natural marshes before entering the river.