Abstract: This study evaluated the effect of combined application of Chinese milk vetch (Astragalus sinicus L.) and different levels of nitrogen (N) fertilizer on dry matter accumulation and N use efficiency of early rice in order to determinate the suitable N level under application of Chinese milk vetch. Field treatments were set with four N application levels (90 kg·hm^-2, 120 kg·hm^-2, 150 kg·hm^-2 and 180 kg·hm^-2) under 22 500 kg·hm^-2 Chinese milk vetch using the winter fallow field with general N fertilizer level (150 kg·hm^-2) as the control. In the experiment, Chinese milk vetch was planted in winter and incorporated into soil in spring before early rice transplant, while N fertilizer was applied as base fertilizer, tillering fertilizer and heading fertilizer with 5:3:2 rate. The rice plants were sampled at tillering, booting, heading, filling and maturity stages of the early rice. Dry matter weight and N content were assayed, and the relative indexes were calculated. The results showed that all N application treatments had higher dry matter accumulation than the control. Specifically, the treatments with Chinese milk vetch combined with 90 kg (N)·hm^-2 and 120 kg (N)·hm^-2 increased dry matter (9.65 t·hm^-2 and 9.97 t·hm^-2) respectively by 11.18% and 14.86% compared to the control. Dry matter accumulation increased with duration of growth and peaked twice, one at sowing-tillering stage (19.26%-24.77%) and the other at heading-filling stage (45.23%-52.75%). The combined application of Chinese milk vetch and N fertilizer supported N integration that was beneficial to rice growth. The amounts of N accumulation of early rice treated with Chinese milk vetch and N fertilizer were higher than the control, with increases of 6.95%-18.68%. Among the treatments, that of Chinese milk vetch combined with 90 kg (N)·hm^-2 had the highest production efficiency of dry matter and N harvest index, with respective increases of 3.94%-14.08% and 6.65%-14.90% over other treatments. The results indicated that combined application of Chinese milk vetch with 90 kg (N)·hm^-2 or 120 kg (N)·hm^-2 was satisfactory because they reduced N fertilizer and increased N use efficiency at the same time, which was the ideal fertilization model for the study area.