The large area and fast development of aeolian sandy lands in the Yarlung Zangbo River basin of Tibet, China has caused immense damages to local social and economic development, and thus, it is necessary to carry out a series of ecological engineering to recover vegetation on moving sand dunes. However, revegetation potentiality of moving sandy land has rarely been reported with integrated studies, and under the harsh conditions of habitat stress, revegetation potentiality of different types of moving sandy lands is still unclear, which is one of most important technical bottlenecks and critical basis for policy making as well as for ecological restoration. In this paper, three types of moving sandy lands (on flood plain, on river bank, on mountain slope, respectively) in alpine valley in the middle reaches of the Yarlung Zangbo River basin were first chosen as the study objects, and then evaluation indicators and criterion systems of revegetation potentiality were established on the ground of the division of vegetation succession stages and the analysis of their habitat characteristics on different moving sandy lands. Thereafter, a comprehensive evaluation model for revegetation potentiality (CEMRP) on moving sandy land in the alpine valley was built using the analytic hierarchy process (AHP) and the fuzzy comprehensive evaluation method. The results showed that the potentiality of moving sandy land on mountain slope was the strongest at Level I and with the biggest membership degree value (0.2884), versus little strong to moving sandy land on flood plain at Level II and with the biggest membership degree value (0.2869), little poor to moving sandy land on river bank at Level IV and with the biggest membership degree value (0.3296). The congenital conditions which affected the habitat of vegetation growth, such as climate (precipitation, temperature, sunshine duration), topographic (elevation, slope, aspect) and soil conditions (moisture, pH value, organic matter, total nitrogen, texture, temperature of sand dune), may determine what the status of vegetation recovery would be. However, vegetation recovery measures which could be treated as acquired conditions, could be used to adjust revegetation potentiality. In addition, by adopting suitable countermeasures (suitable species, fencing, sand-protecting barrier) for artificial revegetation on different types of moving sandy lands, vegetation potentiality could be promoted directly, while the status of vegetation recovery also could be improved through altering soil conditions indirectly. There were three experimental plots established in 2008 for the observation of vegetation recovery. We integrated the vegetation recovery status of different types of moving sandy lands in 2014, and the accuracy and applicability of CEMRP was validated and discussed. The result indicated that CEMRP fitted very well with vegetation growth on different types of moving sandy lands, and could be used as a decision support system for policy makers to identify the difficulties and the priority of ecological rehabilitation for different types of moving sandy lands, even for different regions under different climatic zones in the Tibetan Plateau.