As one of principal paleoclimatology proxy, documentary data have been proved to be of unique importance for reconstructions of climate change in quantitative for past thousands of years. This paper provides an overview of the information of meteorological records derived from Chinese historical documents and the advantages and drawbacks of historical records used as data sources on paleoclimatology reconstruction. Special attention is given to the summary on principal approaches for the reconstruction of climate change in quantitative based on various information recorded in different documentary sources. Furthermore, the reconstructions of temperature and wet/dry changes in central eastern China for the past 2000 years are presented, which were reconstructed by calibration on assembling several separate data derived from different documentary sources and different periods respectively. It aims to provide the methodology for the derivation of proxy data from historical documents in objective and the reconstruction of climate change in high quality. As well as, it will help to improve the approaches for future study on reconstruction of climate parameters and climate changes in higher confident level by using documentary evidences. In brief, there are 4 kinds of meteorological records from historical documents: weather observations, climate extreme and disasters, plant phenology and weather-dependent natural phenomena, descriptions and impacts of local or regional climate. They were recorded in different documents with different format, continuity, as well as the level of details and degree of quantitative descriptions. The approaches of regression analysis, physical model, grade assessment (by an ideal frequency criteria usually) with indices on an ordinal scale, counting (especially for frequency of climate event or disasters), analogy analysis are usually used for derivation and calibration between documentary proxies and climate parameters to reconstruct time series. However, the approaches for the derivation and calibration between documentary proxies and climate parameters in quantitative reconstruction of time series vary with the types of records. For example, the drought/flood grade was usually assessed by an ideal frequency criteria with 10% for severe drought in grade 1, 20% for drought in grade 2, 40% for normal in grade 3, 20% for flood in grade 4,and 10% for heavy flood in grade 5,based on the description of intensity,duration and area of the disaster, as well as its impact respectively. To deal with the discontinuities in original data for reconstructing a time series in consistency, the approaches of regression analysis, variance match, de-trending of the available records are always adopted for calibration on assembling several separate data derived from different documentary sources and different periods respectively. Moreover, it's necessary to formulate more approaches,especially for data interpolation,in future study.