Nitrogen (N) loads from municipal and mine wastewater discharges typically increase N concentrations in recipient water bodies which should get more attention especially in cold-climate regions. This study compared N removal efficiency of six constructed wetlands (CWs) treating mine waters and three CWs polishing municipal wastewater. There were clear impacts of point source N loading to recipient water bodies in all cases studied and >300-fold increase in N was seen in some cases. First-order N removal coefficient was determined for seven of these CWs. All CWs studied were observed to remove N efficiently during the warm growing season but the amount of N released increased significantly during the cold season. Although some year-round purification was achieved by both peat-based and pond-type CWs, removal of nitrate + nitrite-N ((NO3− + NO2−)-N) was low during winter. The first-order N removal coefficient varied from 4.9 · 10−6 to 1.9 · 10−3 d−1 and showed that peat-based CWs were slightly more efficient in N removal than pond-type CWs. However, purification efficiency was steadier and higher for pond-type CWs, as lower hydraulic load or longer water residence time compensated for purification performance. Pond-type CWs showed mean removal efficiency of 59% and 46% for ammonium-N (NH4+-N) and (NO3− + NO2−)-N, respectively, whereas peatland-type CWs had lower removal efficiency for NH4+-N (mean of 26%) and in many cases negative removal for (NO3− + NO2−)-N. Correlation analysis revealed no clear, systematic relationship between temperature and N removal. However, in some CWs the highest correlation was between temperature and (NO3− + NO2−)-N, reflecting lower denitrification rate at lower temperature. More than 50% removal was found to require a hydraulic load below 10 mm d−1. In order to achieve 70% of NH4+-N removal, Ntot load lower than 75 g m−2 year−1 and a residence time longer than 80 d are needed in CWs in cold-climate regions.