BACKGROUND: In Brazil, dengue epidemics erupt sporadically throughout the country and it is unclear if outbreaks may initiate a sustainable transmission cycle. There are few studies evaluating the ability of Brazilian Aedes aegypti populations to transmit dengue virus (DENV). The aim of this study was to compare DENV susceptibility of field-captured Ae. aegypti populations from nine distinct geographic areas of the city of Belo Horizonte in 2009 and 2011. Infection Rate (IR), Vector Competence (VC) and Disseminated Infection Rate (DIR) were determined. METHODS: Aedes aegypti eggs from each region were collected and reared separately in an insectary. Adult females were experimentally infected with DENV-2 and the virus was detected by qPCR in body and head samples. Data were analyzed with the Statistical Package for the Social Sciences version 17. RESULTS: IR varied from 40.0% to 82.5% in 2009 and 60.0% to 100.0% in 2011. VC ranged from 25.0% to 77.5% in 2009 and 25.0% to 80.0% in 2011. DIR oscillated from 68.7% to 100.0% in 2009 and 38.4% to 86.8 in 2011. When the results were evaluated by a logistic model using IR as covariate, North, Barreiro, South-Central and Venda Nova showed the strongest association in 2009. In 2011, a similar association was observed for South-Central, Venda Nova, West and Northeast regions. Using VC as covariate, South-Central and Venda Nova showed the most relevant association in 2009. In 2011, South-Central, Venda Nova and Barreiro presented the greatest revelation associations. When DIR data were analyzed by logistic regression models, Pampulha, South-Central, Venda Nova, West, Northeast and East (2009) as well as South-Central, Venda Nova and West (2011) were the districts showing the strongest associations. CONCLUSIONS: We conclude that Ae. aegypti populations from Belo Horizonte exhibit wide variation in vector competence to transmit dengue. Therefore, vector control strategies should be adapted to the available data for each region. Further analysis should be conducted to better understand the reasons for this large variability in vector competence and how these parameters correlate with epidemiological findings in subsequent years.