The emission of carbon dioxide varies systematically with the rank and type of coal combusted. Hence use of a single default emission factor proposed by IPCC (Inter Governmental Panel on Climate Change) for entire categories coals may not be appropriate option to obtain a reliable estimate of carbon dioxide emission level or towards the preparation of national carbon dioxide inventory. Even predictive equations developed based on the coals of different origin may not work well with coals of a specific origin. Several linear predictive equations were thus developed separately for coking and non-coking coals of Indian origin for the estimation of carbon dioxide emission utilizing basic coal parameters such as VM, FC, GCV and NCV on different basis. Large numbers of authenticated data set were used for multiple regression analysis and good correlations were obtained. Those equations were also validated with different data sets of Indian coals. Its applicability towards estimation of CO2 emission from power plant was also studied and uncertainty in CO2 estimation is revealed. The developed equations may be utilized to get a realistic estimate of carbon dioxide emission with specific cases where Indian coals are mostly used. (c) 2008 Elsevier Ltd. All rights reserved.