Based on a dataset of flowering dates of Japanese Cherry Blossoms (P. yedoensis Mats.) recorded continuously in the last 62 years in Wuhan University Campus and climatic data in the same period, the authors analyze the change in first-flowering and flower fading dates and its correlation with the climatic factors, and establish the statistical models of first-flowering dates with the wintertime temperature. The analysis could not only provide a further evidence for the local climate change, but also formulate a tool for predicting the flowering dates. The results show that: (1) In the time period 1947-2000, the first-flowering dates of Japan Cherry Blossoms advance significantly at a rate of 2.17d per decade, with an overall trend of 11.7d in the 54 years, and the fading dates postpone a little with a linear trend of about 1.83d for the same period. This makes the duration of florescence increases by 13.55d. However, the annual variability of these indices increases obviously at later stage of the period. (2) First-flowering dates have a negative correlation with average temperature of months from December to March, and the negative correlations with average temperature of February and wintertime (December to February) are statistically significant. The wintertime temperature is increasing significantly in recent years, and this is the main reason for the obvious advancement of the first-flowering dates of Japanese Cherry Blossoms. A 1℃ increase in average temperature of February and wintertime will make the first-flowering dates advance by 1.66d and 2.86d respectively. (3) A linear and nonlinear statistical model of the first-flowering dates are built up with the average temperature of February and wintertime, and it has been tested against the 2001-2008 observations, with the mean error being about three days. The nonlinear model has a good performance especially for the years of 2004 and 2007 when the abnormally earlier first-flowering phenomenon occurs.