The effect of nitrogen (N) application on seed yields and yield components in Leymus chinensis (Trin.) Tzvel., a perennial rhizomatous grass, was measured in a field experiment with two saline-sodic soils at Da’an Sodic Land Experiment Station during 2010–11. Two grassland field sites were classified as moderately saline–sodic (MSSL) and severely saline–sodic (SSSL). Application rates of N at each site were 0, 30, 60, 90, 120, 150, 180 and 210 kg ha^–1. Application of N significantly improved seed yield mainly through increased spike number (R² = 0.96, P ≤ 0.001). Compared with nil N, seed yield increased 7.4–10.9 times with N application of 150 kg ha^–1 at MSSL, and 5.3–7.5 times with N application of 120 kg ha^–1 at SSSL. However, absolute increases at SSSL were relatively small. Some significant differences (P ≤ 0.01) in seed yield occurred between 2010 and 2011 with different N application rates in the same soil, and between MSSL and SSSL in the same year. Increasing N application rate significantly decreased N physiological efficiency (NPE) but increased N apparent-recovery fraction (NRF) and N partial-factor productivity (NPP) at both sites. Seed yield and NPP indicated that the optimal N application rates to increase yield were 150 kg ha^–1 at MSSL and 120 kg ha^–1 at SSSL. High soil pH was the major factor adversely impacting seed yield, and pH and soil salinity were major factors negative affecting NPE, NRF and NPP as well as decreasing the positive effect of N application. Nitrogen application is a practical and effective method to increase seed yield of L. chinensis in saline-sodic grasslands of Northeast China, particularly when soil pH and salinity are not limiting.