Surface-flow constructed wetlands (CWs) are potential cost-efficient solutions to mitigate phosphorus (P) loads from agricultural areas to surface waters. Hydraulic and phosphorus loading rates (HLR and PLR) are critical parameters that regulate P retention in these systems. The present study aimed to assess P retention from agricultural drainage discharge by three new CWs in Denmark during three years. The CWs presented different geologies, and received event-driven drainage discharge. The relations of HLR, PLR, dominant P forms, and iron (Fe) inputs to P retention were also investigated. The drainage discharge varied seasonally and average annual HLR ranged from 17 to 60 m yr(-1) for the three CWs. PLR varied concomitantly to HLR with average annual PLR from 2.8 to 13.5 g m(-2) yr(-1), where specific total P (TP) retention followed its variations. Specific TP retention responded differently to HLR and PLR in the CWs, depending on the dominant P forms at the inlet water. The CWs worked as P sinks and presented similar or higher retentions than other Nordic CWs with similar loads. PLR explained most of the variation in specific TP retention (52-72%), whereas HLR only explained 20-34%. Specific TP retention responded more promptly to PLR when the inlet water was dominated by particulate P (PP), since it is more easily retained than dissolved P. TP retention efficiency varied more irregularly and the annual retention ranged from 24 to 66%. It presented weak correlations to HLR and PLR, and depended on the dominant P forms and major retention mechanisms, as well as P biogeochemical stability. Thus, increasing TP retention efficiency occurred under higher stability of sediment bound P, as supported by higher total Fe:PP molar ratios at the inlet water. The supply of P sorbents by drainage discharge, soil redox conditions and Fe:P molar ratios are considered key parameters controlling the biogeochemical stability of P, and thus the long-term P retention performance in CWs.