High soil nitrate-N accumulation has been observed in North China Plain (NCP), but it was seldom considered as a N source in N management due to the lack of data on crop response to soil nitrate-N accumulation. A total of 124 on-farm N-response experiments were conducted from 2003 to 2006 in seven key winter wheat (Triticum aestivum L.) production regions of NCP to evaluate wheat yield response to residual soil nitrate-N. High soil nitrate-N accumulation ( ≥172 kg N ha-1 in 90 cm soil depth) was observed during wheat growing season. Compared with 165 nitrate-N ha-1 in 0 to 90 soil layer before sowing, added fertilizer N did not increase wheat yield under 262 kg nitrate-N ha-1 condition, but increased residual nitrate-N and N losses and reduced recovery N efficiency (P < 0.05). Wheat yield response to residual soil nitrate-N could be fitted by a linear plus plateau model, and calculated minimum soil nitrate-N content for maximum grain yield was 211 kg N ha-1 (R2 = 0.31) before sowing across all 223 zero-N plots in 124 sites. When two growth periods (before and after shooting stage) were analyzed separately, the coefficient of determination between yield and soil nitrate-N increased significantly. We conclude that winter wheat yield response to added N fertilizer was unlikely if soil nitrate-N content exceeded 72 (0-30 cm soil depth) and 175 kg N ha-1 (0-90 cm soil depth) before sowing and shooting stage, respectively. According to these thresholds, 55% and 37% of farmer's fields did not need to apply N fertilizer before sowing and at shooting stage, respectively.