There have been relatively few attempts to model the distributions of the fruit flies Ceratitis capitata (Wiedemann), Ceratitis cosyra (Walker), and Ceratitis rosa Karsch in China, but the geographic distributions of these species are of considerable concern in terms of biosecurity. In this study, two different modeling methods (genetic algorithm for rule-set prediction [GARP] and maximum entropy species distribution modeling [Maxent]) were used to predict the potential distributions of these three fly species in China, by using distribution records and a set of environmental predictor variables. The results showed that Maxent performed well, compared with modeling by GARP, at each test threshold. For all three species, the results predicted by Maxent agreed with the observed distributions in Africa and in other parts of the world. In China, C. capitata seems to have the highest number of favorable habitat areas, relative to C. cosyra and C. rosa, i.e., Yunnan, Guizhou, Guangxi, Guangdong, Hainan, Fujian, Sichuan and Chongqing, whereas C. cosyra has the smallest range of suitable areas, i.e., Yunnan, some parts of Hainan and Sichuan. The suitable areas for C. rosa are mainly restricted to Yunnan, Hainan, southern Guangdong, and a few areas of Sichuan. The indications are that on the whole, Southwest and South China are the areas with the highest risk for establishment from these three fly species. Jackknife tests reveal that environmental variables associated with temperature have the strongest influence on the potential distributions of all three species relative to other variables.