TM4-O-08 Climate change has influenced the occurrence and distribution of vector-borne diseases in the world. Identifying vulnerable population in response to future changes of climate is therefore important for public health adaptation decision. Our study therefore conducted spatial analysis to examine the linkage between temperature and other environmental factors on dengue fever distribution and predicting the area with potential risk of dengue fever endemics in future climatic change. We used Geographical Information System (ArcGIS 8.2) to demonstrate the spatial patterns of dengue fever occurrence, climatic and nonclimatic factors. The basic spatial unit used in this study is township, and 357 townships with complete case number of dengue fever, weather, population density, and vector surveillance data were included in the analysis. Computerized database containing daily registration of dengue fever cases in Taiwan for the period of 1998 to 2002 were obtained from Taiwan CDC. Logistic regressions were further adopted to evaluate the linkage between spatial characteristics of climatic and nonclimatic factors, and occurrence of dengue fever to further identify risk factors of dengue fever distribution. Incidence rates per square kilometer were used to map out the incidence rate from 1998 to 2002 in each township. Significantly higher monthly average temperature could be observed in those townships with occurrence of dengue fever (nonparametric Mann-Whitney test). Significantly elevated risk of dengue fever occurrence could be observed, with odds ratios between 1.65 and 19.18, when the average monthly temperature for the specific township exceeded certain levels. Warmer winter season seemed to impose relatively higher risk than the warmer summer season. The recovery rate of household vectors higher than 30% in the township was also found to be associated with dengue fever occurrence (OR = 1.80; 95% CI, 1.05–3.02). However, greater population density did not seem to be a critical factor for increasing vector population in Taiwan. The map of risk scores, based on temperature and recovery rate of vector density over 12 months, were plot to indicate the areas with potential risk for dengue fever occurrence. In our study, 39 townships with scores higher than 10, account for almost 11% of 357 townships in Taiwan, have contributed to nearly 95% indigenous cases between 1998 and 2002. The predicted risk scores using temperature and recovery rate of vector density could be considered a worth-taking approach for identifying vulnerable regions in Taiwan to highlight those areas with greater potential risk for dengue fever outbreak in the future.