Abstract: Abstract: The ridge and furrow rainfall harvesting system is known to be an effective planting method for improving rainwater utilization, but suitable fertilizer application rates for foxtail millet under RFRH planting have not yet been determined. In order to clarify the effects of fertilization under RFRH system and excavate its potential to enhance crop yield, we conducted field experiments during 2012-2015 at the Dryland Agricultural Experiment Station, Pengyang city, Ningxia, at the Loess Plateau, China. The experimental site was located at a longitude of 35?79?E and latitude of 106?48?N, at an elevation of 1658 m above sea level. The field experiment included 2 planting patterns (R, ridge and furrow rainfall harvesting planting; T, traditional flat planting) and 4 fertilizer level (F0, no fertilizer; F1, N:P2O5 at 93:48 kg/hm2; F2, N:P2O5 at 186:96 kg/hm2, F3, N:P2O5 at 279:144 kg/hm2), the R model used ridge and furrow widths of 60 cm, a ridge height of 15 cm, and the ridges were mulched with plastic film. The soil moisture of 0-20 cm profile was measured using a manual soil ferric auger at sowing, heading and harvesting stage, soil water storage and crop water consumption were calculated. The dry matter at different growth stages, yield and its components, use efficiency of water and fertilizer, and economic profit were measured. The 4-year experiment results showed that fertilization generally had a significantly positive effect on plant growth, yield formation and water use efficiency. As the increase of fertilizer input rate, the dry matter per plant, grain yield, WUE and economic profit under R planting were improving, but there was no significant increase when fertilizer rate beyond F2 in which the agronomy efficiency decreased, which indicated F2 was a suitable fertilizer input level for R planting. Compare to no fertilizer, the grain yield of the R treatment significantly improved by 28.1%-85.3%, respectively. Compared to traditional flat planting, the R planting could increase yield by 8.3% and 7.8% in the normal year, but decrease yield by 6.1% in the wet year (2012). The yield-increasing effect was best in the medium and low fertilizer treatments. In addition, the R treatment could reduce water consumption than the T treatment. The water consumption of R treatment was 18.3-26.8 mm lower than the T treatment in 2012-2015. In the normal year, the different planting model affected significantly the grain number per ear and harvest index, but did not affect the 1000-seeds weight. The rain number per ear and harvest index of the R treatment was 8.4% and 4.2% higher than the T treatment (P<0.05). In wet year, the planting model did not significantly affect these yield components of millet. Compared with the T treatment, the R treatment significantly increased the water use efficiency of millet in the normal year. The increase under the middle fertilizer treatment was 14.6% and 15.1%. The R treatment could increase agronomy efficiency than the T treatment, which increased with reduction of fertilizer application amount. This effect was better in the normal year than in the wet year. Among all the treatments, the medium fertilizer application in the R treatment in the normal year increased the yield by 7.6%, water use efficiency by 14.9%, agronomy efficiency by 9.9% and economic profit by 6.2% than the T treatment. Therefore, the RFRH system combine with F2 (N: P2O5 at 186:96 kg/hm2) was a reasonable and efficient model for foxtail millet cultivation in semi-arid areas.