Tourism is a rapidly increasing industry and has recently attracted attention as an important contributor to climate change through greenhouse gas emissions. Recently, researchers have focused on the impact of carbon emissions from tourism. Among the greenhouse gases, CO₂ is the main contributor to global warming. Previous energy consumption studies have indicated that CO₂ emissions from tourism can be categorized into three sections, namely transportation, accommodation and activities. Among these sections, tourism transportation accounts for the most significant proportion of the energy consumption and CO₂ emissions. This paper focuses on three scenic spots, the Jiuzhaigou Scenic Area, Xi'an Beilin Museum and Nanjing Pearl Spring Resort, and applies a bottom-up approach to determine the CO₂ emissions related to tourism transportation in these areas in 2010. We investigated the visitors to these three scenic spots by questionnaires using stratified sample method, collected 1404 available samples, and estimated the total amount of carbon emissions in each scenic spot. The results indicated that CO₂ emissions from aviation are higher than those of all other transportation methods. The total CO₂ emissions at the Jiuzhaigou Scenic Area, Xi'an Beilin Museum and Pearl Spring Resort were 386.27, 215.71 and 10.45 Gg, respectively. The higher carbon emissions at Jiuzhaigou were attributed to the spot's long average travel distances. To analyze the mechanism of CO₂ emissions from tourism transportation more thoroughly, we used the carbon cumulative curve to describe the spatial structure of the three scenic spots. The carbon cumulative curve analysis suggested that the average travel distance was the most important factor controlling CO₂ emissions, influencing not only the balance structure, but also the spatial structure of CO₂ emissions from tourism transportation. The tourist attractions with low travel distances had the most unbalanced structure of carbon emissions. Also, the spatial structure of carbon emissions in different subsections of travel distance had significant discrepancies. Finally, we completed a scenario analysis to explore policies to reduce CO₂ emissions from tourism transportation. When we compared four different traffic scenarios, we found that the tourist attractions with high and moderate average travel distances were sensitive to a reduction policy that decreased the proportion of emissions from aviation. Therefore, low-carbon travel modes such as the high speed railway could be promoted as an alternative to aviation, and economic compensation such as discounted tickets could be provided for low-carbon tourists to reduce CO₂ emissions. In contrast, tourist attractions with low average travel distances are sensitive to the policy of switching from private cars to public vehicles. Consequently, we can implement reduction policies like encouraging tourists to travel by bus or bicycle and using carpools to decrease the average load factor. These results have important implications for tourism management departments to understand the structure of CO₂ emissions and establish targeted policies to reduce CO₂ emission from tourist transportation.