Abstract: In the mountainous area of Ansai, the frost hazard occurrence at apple initial flowering stage is frequent, and the prediction of the initial flowering stage of apple is carried out to provide guidance for agricultural production and reduce disaster losses. By using the derivation method of linear regression equation, we analyzed the biological zero, active accumulated temperature and effective accumulated temperature at flowering stage of apple in Ansai mountainous area, explored the correlation between the accumulated temperature of each 10-day period 50 days before the initial flowering and the initial flowering stage, and tested and predicted the initial flowering day by using the regression equation of the relevant significant accumulated temperature. The results showed that: the 10-day average accumulated temperature and maximum accumulated temperature in the 4 10- day periods before the initial flowering stage were significantly correlated with the initial flowering day; and the biological zero of 10-day average temperature was 2.9℃; according to the test, the prediction accuracy rate of the initial flowering day in 2006-2015 was relatively high; the forecast value of the initial flowering day in 2018 was consistent with the actual observed value. The regression equation of the average accumulated temperature and maximum accumulated temperature in the 4 10-day periods before the initial flowering stage of apple can be applied to the forecast of initial flowering stage in future, which can provide a basis for the prevention of meteorological disasters in the flowering period of apple.