Abstract: To provide a methodological support for accurately predicting the initial flowering period of rape in alpine regions. The initial flowering data of the rape and ground meteorological observation data were adopted in this paper over the Guinan County Meteorological Bureau of Qinghai Province from 1991 to 2017, and applied climate propensity rate and statistical analysis such as Mann-Kendall mutation method, studied the relationship between the initial flowering period of rape and meteorological conditions, and predicted it with multiple regression analysis. The results showed that the initial flowering period of rape in the high-cold southern Guinan region was delayed by 3d/10a, and it occurred a mutation in the initial flowering period of rape around 1998 with the warming of the climate. The initial flowering period of rape was significantly positively correlated with the average temperature and ≥0°C of the accumulated temperature in May, and with the average maximum temperature and average wind speed significantly extremely. However, it was significant negative correlation with the precipitation and relative humidity extremely in May. It established a prediction model for the initial flowering period of rape by multivariate regression analysis method. The established prediction model was used to measure the initial flowering period of rape, and the retro-generation effects was good.