Improving nitrogen use efficiency(NUE) and decreasing N loss are critical to sustainable agriculture. The objective of this research was to investigate the effect of various fertilization regimes on yield, NUE, N agronomic efficiency(NAE) and N loss in long-term(16- or 24-yr) experiments carried out at three rice-wheat rotation sites(Chongqing, Suining and Wuchang) in subtropical China. Three treatments were examined: sole chemical N, N+phosphorus(NP), and NP+potassium(NPK) fertilizations. Grain yields at three sites were significantly increased by 9.3-81.6%(rice) and 54.5-93.8%(wheat) under NP compared with N alone, 1.7-9.8%(rice) and 0-17.6%(wheat) with NPK compared with NP. Compared to NP, NUE significantly increased for wheat at Chongqing(9.3%) and Wuchang(11.8%), but not at Suining, China. No changes in NUE were observed in rice between NP and NPK at all three sites. The rice-wheat rotation’s NAE was 3.3 kg kg-1 higher under NPK than under NP at Chongqing, while NAE was similar for NP and NPK at Suining and Wuchang. We estimated that an uptake increase of 1.0 kg N ha-1 would increase 40 kg rice and 30 kg wheat ha-1. Nitrogen loss/input ratios were ~60, ~40 or ~30% under N, NP or NPK at three sites, indicating significant decrease of N loss by P or PK additions. We attribute part of the increase in NUE soil N accumulation which significantly increased by 25-55 kg ha-1 yr-1 under NPK at three sites, whereas by 35 kg ha-1 yr-1 under NP at Chongqing only. This paper illustrates that apply P and K to wheat, and reduce K application to rice is an effective nutrient management strategy for both the NUE improvement and N losses reduction in China. Improving nitrogen use efficiency (NUE) and decreasing N loss are critical to sustainable agriculture. The objective of this research was to investigate the effect of various fertilization regimes on yield, NUE, N agronomic efficiency (NAE) and N loss in long-term 16- or 24-yr) experiments carried out at three rice-wheat rotation sites (Chongqing, Suining and Wuchang) in subtropical China. Three treatments were examined: sole chemical N, N + phosphorus ) fertilizations. Grain yields at three sites were significantly increased by 9.3-81.6% (rice) and 54.5-93.8% (wheat) under NP compared with N alone, 1.7-9.8% (rice) and 0-17.6% NPK compared with NP. Compared to NP, NUE significantly increased for wheat at Chongqing (9.3%) and Wuchang (11.8%), but not at Suining, China. No changes in NUE were observed in rice between NP and NPK at all three sites The rice-wheat rotation’s NAE was 3.3 kg kg-1 higher under NPK than under NP at Chongqing, while NAE was similar fo r NP and NPK at Suining and Wuchang. We estimated that an uptake increase of 1.0 kg N ha-1 would increase 40 kg rice and 30 kg wheat ha-1. Nitrogen loss / input ratios were ~ 60, ~ 40 or ~ 30% under N, NP or NPK at three sites, indicating significant decrease of N loss by P or PK additions. We attribute part of the increase in NUE soil N accumulation which increased increased by 25-55 kg ha-1 yr-1 under NPK at three sites, while by 35 kg ha-1 yr-1 under NP at Chongqing only. This paper shows that apply P and K to wheat, and reduce K application to rice is an effective nutrient management strategy for both the NUE improvement and N losses reduction in China.