A 2-year field microplot experiment was conducted to determine the effects of nitrogen (N) splits on grain yields and the fate of 15N-labelled fertilizer applied to plastic mulched maize (Zea mays L.). Three N split applications at the same rate of 225 kg N hm−2 were performed. The N was applied at the day before sowing, the eight-leaf stage (V8), and the silking stage (R1) in the following ratios: 100 %–0–0 (N1), 40–60 %–0 (N2), and 40–30–30 % (N3). 15N-labelled urea (10.14 atom%) was used to trace the fate of each N application in the microplots. The results showed that grain yields increased by 8.3 and 10.7 % in the treatments N2 and N3, respectively, compared with the N1 treatment. Plant N uptake derived from fertilizer (Ndff, %) averaged 26.8–32.4 % compared with 67.6–73.2 % derived from soils (Ndfs, %). Split applications of N significantly increased the Ndff in plant. The residual 15N in the 0–200 cm soil layer ranged from 48.3 to 51.3 % at maize harvest, approximately half of which remained in 0–20 cm soil layer. The 15N recovery efficiency (15NRE) was 37.5 and 39.1 % for treatments N2 and N3, respectively, and was significantly higher than that for N1 treatment (27.9 %). The potential N losses in the treatments N2 and N3 were 11.2 and 12.7 %, respectively, and were significantly lower than losses in treatment N1 (22.2 %). In conclusion, applying N with two splits could produce higher grain yields, higher NRE, and lower N losses in semiarid plastic mulched maize cropping system.