A constructed wetland composed of a pond- and a marsh-type wetland was employed to remove nitrogen (N) and phosphorus (P) from effluent of a secondary wastewater treatment plant in Korea. Nutrient concentrations in inflow water and outflow water were monitored around 50 times over a 1-year period. To simulate N and P dynamics in a pond- and a marsh-type wetland, mesocosm experiments were conducted. In the field monitoring, ammonium (NH4+) decreased from 4.6 to 1.7 mg L−1, nitrate (NO3−) decreased from 6.8 to 5.3 mg L−1, total N (TN) decreased from 14.6 to 10.1 mg L−1, and total P (TP) decreased from 1.6 to 1.1 mg L−1. Average removal efficiencies (loading basis) for NO3−, NH4+, TN, and TP were over 70%. Of the environmental variables we considered, water temperature exhibited significant positive correlations with removal rates for the nutrients except for NH4+. Results from mesocosm experiments indicated that NH4+ was removed similarly in both pond- and marsh-type mesocosms within 1 day, but that NO3− was removed more efficiently in marsh-type mesocosms, which required a longer retention time (2–4 days). Phosphorus was significantly removed similarly in both pond- and marsh-type mesocosms within 1 day. Based on the results, we infer that wetland system composed of a pond- and a marsh-type wetland consecutively can enhance nutrient removal efficiency compared with mono-type wetland. The reason is that removal of NH4+ and P can be maximized in the pond while NO3− requiring longer retention time can be removed through both pond and marsh. Overall results of this study suggest that a constructed wetland composed of a pond- and a marsh-type wetland is highly effective for the removal of N and P from effluents of a secondary wastewater treatment plant.