The Northeast China (NEC) accounts for more than 30% of the national total maize planted area (grain yield). Adjusting the sowing dates has been considered an effective measure to adapt to climate change, but there was little evidence that how well the producers had done. In this study, we used phenology observations at 67 stations from 1981 to 2014 to detect trends in actual sowing dates, then the agricultural production systems simulator, APSIM-Maize model, was used to assess the effects of changes in observed sowing dates on maize phenology and yields. During the past 34-year period, the actual maize sowing dates show a delaying tendency, at a rate of 1–6 days per decade, but there are significant fluctuations among years. For per day delay in the sowing dates, the whole growing season was shortened by 0.1%. Delaying sowing dates reduced the solar radiation interception during the vegetative period as well as the thermal time during the reproductive period. As a result, the overall maize potential yield was negatively affected in NEC; for per day delay in the sowing dates, the potential yield was decreased by 0.6%. By contrast, advancing sowing dates in some years led to increases in both the solar radiation interception during the vegetative period and the thermal time during the reproductive period. However, these increases showed various effects on the maize potential yield across different parts of the study region. We detected a positive effect of advancing sowing dates on maize potential yield in the high latitudes, at a rate of up to 1.6%. By contrast, in the low latitudes, the negative effect of advancing sowing dates on maize potential yield was dominant, at a rate of up to 2.7%.