Climate change research in South China extending back before the current century not only has considerable importance to researching the features of climate change and predicting future climate change and climate disasters in South China, but also plays a positive role in analysis of the characteristics of climate change throughout China. In this paper, there are two data sources: historical documents and modern meteorological observations. Historical documents include Yu- Xue- Fen- Cun records and collections from local gazettes and other historical data. A total of 1499 records from stations operational from 1736 to 1911 were collected from Yu- Xue- Fen- Cun records with about 9 stations recorded yearly, and 1237 records from 1710 to 2000 were collected from other historical documents with about 4 records yearly. Based on the features of records in historical documents and characteristics of winter climate in South China, methods for reconstructing the winter cold index were established to reconstruct the series of winter cold index for analysis of the characteristics of cold and warm in winters in South China from 1710 to 2009 by clustering analysis, the Mann-Kendall method and power spectrum analysis. The results showed the following: (1) The average winter cold index in South China was 2.2 from 1710 to 2009,with a maximum index range of neighboring years of 3.0 (the difference between 1834 and 1835) and maximum index difference of adjacent decades of 0.7 (the difference between the 1890s and 1900s), which indicated an obvious change in winter cold. (2) Winter temperatures were warmer in the 20th century than in the 19th century. (3) The cold and warm changes in winters in South China from 1710 to 2009 could be divided into three stages: warmer winter existed during the two periods from 1710 to 1834 and from 1893 to 2009, while it colder winters were observed in from 1835 to 1892. (4) The cold winter index in South China had no apparent change from 1710 to 2009, with an obvious warming trend indicated by the Mann- Kendall method. (5) The results of spectral analysis showed the cold winter index in South China had dominant cycle of 100,50, 40,and 8 years. (6) There was a significant correlation between winter cold index and instrumental series in the same periods, which showed that the constructed series could reflect the features of winter change in the past 300 years at some scale and had some reference meaning regarding research into the characteristics of winter of South China and China overall.