As an industry with high carbon emissions, tourism’s carbon emissions have always been an issue that needs to be paid attention to on the road to carbon reduction. This study proposes an input-output analysis-based index decomposition (IOID) method to predict carbon emission of tourism sector is proposed through integrating the input-output analysis (IOA), Logarithmic Mean Divisia Index (LMDI), and STIRPAT methods into a general framework. The method has advantages of synergistically predicting the direct and indirect carbon emissions of an economy or specific sectors. Then, the IOID method is applied to Fujian Province (in China) as a case study. The major findings are (i) The carbon emissions of the tourism industry in Fujian Province are growing very rapidly, total carbon emissions in 2017 were 21.46Mt, 3.07 times increase from 6.99Mt in 2007. (ii) The number of tourists is the biggest factor influencing the carbon emissions of the tourism industry. In order to curb carbon emissions from the tourism industry, consider controlling the size of tourists. (iii) Combining the STIRPAT method, the carbon emissions of Fujian Province in the future were predicted, and two optimal development scenarios were selected that can both make Fujian Province reach its carbon emissions peak before 2030 and meet the sustainable and healthy development of the economy.