Abstract: Northeast China is the largest spring maize production area in China and plays a vital role in ensuring food security. Temperature is an important environmental factor affecting agricultural production, especially for mid-high latitudes. Accumulated temperature, as a measure of heat, can be used to estimate the growth rate of crops, and the advance or delay of the growth period will affect the accumulation of dry matter in crops. Therefore, accurate forecast of maize growth period can promote current farming systems and management measures to ensure spring maize yield. As one of the most commonly used accumulated temperature calculation methods, the active accumulated temperature is refered to the accumulation of the average daily temperature over a period of time above a certain threshold, which is widely used in phenological period forecasting, agrometeorological disaster assessment, introduction of new varieties, and agro-climatic thematic analysis and zoning. The active accumulated temperature required for the growth period of the crop is not a constant. The relationship between crop development speed and temperature is not linear. Affected by the crop variety and environmental factors, the active accumulated temperature reflects the instability to influence application effect. Therefore, it is of great significance to modify the existing accumulated temperature models and improve the stability of accumulated temperature for better application. Based on the growth and development of spring maize, 5 agrometeorological stations in Northeast China, Hailun, Dunhua, Changling, Kuandian and Zhuanghe are selected to comprehensively analyze the meteorological factors affecting the stability of accumulated temperature and to revise the widely used active accumulated temperature model. After evaluating its effect, the revised model is applied to the growth period forecast of spring maize. Results show that due to its important role in affecting the stability of the accumulated temperature, the temperature is the key factor considered in the model revision. The revised model improves its stability and reduces variation coefficients in the emergence-heading period and the heading-maturation period by 0.42% and 1.42%, respectively. Using data in 1981-2010 for hindcast and data in 2011-2017 for forecast test, compared with the original active accumulated temperature model, the forecast error in revised model during the mature period is reduced by 3.78 d and 1.1 d. The revised model does not improve the forecast of the heading period.