Background: Because of the urbanization and the lack of basic mosquito control facilities, constructions sites might increase the transmission risk of dengue. For example, the outbreak in Zhanjiang Prefecture, China, 2018, started at a construction site, then spread to surrounding communities. This study aims to investigate how to effectively reduce the dengue transmission risk related to construction sites.Methods: The study was based on an outbreak of dengue fever (DF) occurred in Zhanjiang Prefecture, Guangdong Province in 2018. We simulated the transmission of dengue virus between human and mosquitoes at the construction site and in the neighboring community separately, as well between them by assuming human movement between the construction site and the neighboring community. Susceptible-Exposed-Infectious/Asymptomatic-Recovered (SEIAR) model was used for the human, while SEI model was used for the mosquitoes. We obtained the transmissibility parameters by fitting the model to observed number of incident cases before local interventions started and the coefficient of determination (R2) was used to quantify the significance of the fit. We then evaluated the effectiveness of different intervention scenarios targeting at reducing the transmissibility between different human and mosquito subpopulations at different locations (i.e. construction sites and community) quantified by the total attack rate (TAR) and duration of the outbreak (DO).Results: A total of 467 cases of dengue fever were reported in Zhanjiang Prefecture in 2018, among which 102 were located at a construction site and 131 were located in the surrounding community. The values of R2 were 0.829 (P < 0.001) and 0.878 (P < 0.001) for the construction site and the community, respectively. The modelling outcome indicated that without interventions, the number of cases on the construction site would reach to 156, yielding a total attack rate (TAR) of 31.25% (95% confidence interval [CI]: 27.18%- 35.31%). In the community, the number of the cases would be much more than the reported data and up to 10796, yielding a TAR of 21.59% (95%CI: 21.23%- 21.95%). When the transmission route from mosquitoes to people is cut off in the community, the number of cases in the community would decrease to a minimum of 33 compared with other situations we simulated, yielding a TAR of 0.068% (95%CI: 0.05%- 0.09%) and a duration of outbreak (DO) of 60 days. When the transmission route from infectious mosquitoes in the community and the construction site to susceptible people on the construction site is cut off at the same time, the number of cases in the construction site would drop to a minimum of 74, yielding a TAR of 14.88% (95%CI: 11.76%- 18.00%) and a DO of 66 days.Conclusions: Construction sites may facilitate the dengue outbreaks into neighboring communities, SEIAR-SEI model can be efficiently applied to simulate outbreaks of dengue fever. Cutting off the transmission routes and properly controlling vectors in communities is an effective measure to control this disease.