Abstract: The paper aims to clarify the seasonal variation characteristics under the background of global warming. The authors adopted correlation analysis and mutation test methods to study the beginning date and length variation characteristics of the four seasons and their relationship with temperature changes based on daily temperature data of Beijing Observatory during 1951-2014. The results showed that: (1) the perennial average beginning date of spring, summer, autumn and winter was April 3, May 26, September 16, November 3, respectively; there were a significant advancement in the beginning dates of spring and summer and a prominent delay in the beginning dates of autumn and winter in Beijing during 1951-2014; (2) winter was the longest season followed by summer and spring, while autumn was the shortest season; the summer in Beijing was prolonged during 1951-2014 which was influenced by the advancing of the beginning date of summer and the delaying of the beginning date of autumn, while the winter significantly was shortened which was influenced by the advancing of the beginning date of spring and the delaying of the beginning date of winter; (3) the annual average temperature in Beijing increased significantly with a linear-change-rate of 0.35℃/10a during 1951- 2014; it showed that the temperature change in Beijing had a significant effect on climate seasonal variation, when temperature increased, the beginning dates of spring and summer advanced, the beginning date of autumn delayed, overall the summer was prolonged; (4) the beginning date and length of the four seasons in Beijing had significant mutations around 1990. In conclusion, the temperature, the beginning date and the length of the four seasons in Beijing have a disequilibrium response to the climate warming.