This study examines the key properties of tourist destination networks. We constructed a network data set using a large collection of tourist trip data that contains 217,686 distinct trips taken by 200,153 tourists across 39,901 cities in 195 countries. Using the data set we mapped and simulated the tourist destination networks using three canonical network models, namely the Erdos–Rényi network, Watts–Strogatz network, and the Barabási–Albert network. We found that the destination networks exhibit long diameters, in contrast to many social and economic networks, but conform to the high clustering and scale-free properties. We found that none of the three network models can adequately capture the formation mechanisms of the destination networks; the destination networks share certain, but not all, of the properties of each of the three network models. Such inadequacy calls for developing new models to study the network formation of tourist destinations.