Abstract: ObjectiveTo explore the association of Breteau index（BI）, imported cases, and meteorological factors with local epidemic of dengue fever, so as to propose early warning thresholds.MethodsGuangzhou was selected as study site and years 2002-2013, as study period. The classification and regression trees model was applied to analyze the associations between the predictive variables including average monthly BI, average monthly imported cases, average monthly temperature,monthly maximum temperature,monthly minimum temperature,average monthly temperature difference,monthly total rainfall, average monthly rainfall,average monthly relative humidity, average monthly atmospheric pressure, and the outcome variables consisting of local dengue whether or not occur and the number of indigenous cases at month level separately.ResultsA total of 3 996dengue cases, including 3 769 (94.32%) indigenous cases and 227 (5.68%) imported cases, were reported in 44 (30.56% of 144) months during the period of 2002 - 2013. In the classification tree model for whether or not the local cases of dengue fever would appear, the number of imported cases, BI, maximum temperature and minimum temperature of the months were eventually incorporated. When the average monthly imported cases were ≥3.5, the RR was 3.00 (2.22 - 4.05) indicating the risk for local cases to occur was the greatest; when the average monthly imported cases were <3.5 and average monthly BI was ≥8.59, the RR was 2.40 (1.62 - 3.55); when inputting the average monthly imported cases <3.5, average monthly BI <8.59, monthly maximum temperature ≥31.41 ℃, and monthly minimum temperature <24.90 ℃, the RR was 2.18 (1.29 - 3.68). In the regression tree model for the severity of indigenous cases, average monthly BI, average monthly temperature, and average monthly imported cases were incorporated. When the average monthly BI was ≥5.29 and average monthly temperature was <27.04 ℃, the RR was the highest (5.11，3.36 - 7.77) ; when the average monthly temperature was ≥27.04 ℃ and everage monthly BI was ≥9.16, the RR was 3.20 (2.06 - 4.96); when the average monthly BI was <5.29 and the average monthly imported cases were ≥3.5, the RR was 2.22 (1.40 - 3.53).ConclusionThe everage monthly imported cases and everage monthly BI are most important two factors for whether or not local dengue cases would appear, and the average monthly BI and monthly average temperature, most important two factors for the severity of local dengue epidemic.