Few systematic studies have been carried out on integrated N balance in extremely arid oasis agricultural areas. A two-year field experiment was conducted to evaluate the N input and output balances under long-term fertilization conditions. Five treatments were chosen, namely CK (no fertilizer), NPK, NPKS (10% straw return N and 90% chemical N), NPKM (one third urea-N, two thirds sheep manure) and NPKM+ (1.5 times NPKM). The results show an abundance of dry and wet N deposition (33 kg N ha-1 yr-1) in this area. All treatments (excluding CK) showed no significant difference in wheat production (P>0.05). NPKM gave higher cotton yields (P<0.05). In both crops, NPKM and NPKS treatments had a relatively higher N harvest index (NHI). 15N-labeled results reveal that the fertilizer N in all N treatments leached to<1 m depth and a high proportion of fertilizer-N remained in the top 60 cm of the soil profile. The NPKM+ treatment had the highest residual soil mineral N (Nmin, 558 kg Nd ha-1), and NPKM and NPKS treatments had relatively low soil Nmin values (275 and 293 kg N ha-1, respectively). Most of the treatments exhibited very high apparent N losses, especially the NPKM+ treatment (369kg N ha-1). Our arid research area had a strikingly high N loss compared to less arid agricultural areas. Nitrogen inputs therefore need careful reconsideration, especially the initial soil Nmin, fertilizer N inputs, dry and wet deposition, and appropriate organic and straw inputs which are all factors that must be taken into account under very arid conditions.