Combining a strong tourism economic performance and less environmental effect is one of the important objectives of current Chinese tourism development. As a total factor productivity index of green tourism, carbon dioxide emissions efficiency of tourism has become an important tool which measures the relationship between carbon dioxide emission of tourism industry and tourism economic growth. Based on the panel data about provincial input-output of tourism industry, the carbon dioxide emissions efficiency of tourism of 30 provinces in mainland China is measured by the Slack-Based Measure (SBM) considering the unexpected output. In addition, the exploratory spatial data analysis model (ESDA) is applied to analyze the spatial distribution characteristics of carbon dioxide emissions efficiency of tourism. At last, the paper analyzes the influencing factors, and their spatial-temporal heterogeneity through the geographical weighted regression model (GWR). The key results are as follows:(1)Carbon dioxide emissions efficiency of tourism in China is low, but recorded obvious progress over the observed period. A low average carbon dioxide emissions efficiency of tourism of 0.38 was calculated in 2001, increasing to almost 0.47 in 2011. The carbon dioxide emissions efficiency of tourism of only a few provinces achieves the best production frontier, while most of provinces need to optimize the input-output of tourism.(2)Spatial distribution characteristics are also significant. The carbon dioxide emissions efficiency of tourism has significant characteristics in spatial correlation which are still strengthening constantly. Local spatial pattern, which initially formed the spatial structure with Shanghai as the center of the high value agglomeration regions and provinces in Northwest China as the center of low concentrated area, is relatively stable, and has not changed substantively over the observed period.(3)Carbon dioxide emissions efficiency of tourism is affected by a combination of factors. Tourism economic scale, technology effect, the level of urbanization, property right structure of tourism and industrial structure on carbon dioxide emissions efficiency of tourism have a spatial-temporal heterogeneity, which causes the migration and accumulation of dioxide emissions efficiency of tourism.(4)The spatial-temporal heterogeneity are recognized by geographical weighted regression model. The tourism economy scale has an obvious promoting effect on carbon dioxide emissions efficiency of tourism, especially in the Central and Western area. The promoting effect of urbanization gradually is weakening, and with the rapid development of urbanization, the inhibition effect exists in most provinces. The technology effect has also an obvious promoting effect, and the high value region concentrates from the Central and Eastern area to the North and Northeast of China. The property right structure of tourism exhibits a clear south-north division, while it has an obvious promoting effect in Southern and an obvious inhibition effect in Northern China. The industrial structure effect has an obvious promoting effect in the Southwest and an obvious inhibition effect in the North and Northeast of China. These results provide a theoretical basis for implementation of differentiation strategy in order to promote carbon dioxide emissions efficiency of tourism in China.