Abstract: In order to meet demands of comprehensive weather observation in modern meteorological service, a development and assessment program of new automatic weather station (NAWS) is sponsored and launched by Meteorological Observation Center of China Meteorological Administration. Besides the traditional louver temperature observation system (LTS), NAWS can also be equipped with aspirated radiation shield temperature observation system (ASPTS), which borrows the design experience of US Climate Reference Network, for the purpose of achieving future long term homogeneous temperature observations. NAWS has been employed in the surface meteorological station in some provinces of China. Differences between LTS and ASPTS results, influencing factors and the correction method all need investigation, therefore, a parallel experimental measurements consists of LTS and ASPTS is conducted. The experiment is carried out in Nanjing University of Information Science & Technology (32°12′N, 118°42′E, elevation is 32 m) from August 2009 to July 2010. Maximum and minimum temperature measurements derived from LTS and ASPTS are compared, and biases of extreme values and difference of the occurrence times are examined under different regimes of ambience wind speeds. A correction model based on ambient wind speed is developed and checked. Results indicate that differences of daily extreme temperature between LTS and ASPTS are not subject to the normal distributions, while they demonstrate a right skewed state and a great degree of deflection. The concordance rate of maximum temperature between LTS and ASPTS is 90.0%, while that of minimum temperature is 81.5%. The gross error rate of maximum temperatures between LTS and ASPTS is almost the same as that of minimum temperature, both of which are about 3.0%. Compared with ASPTS measurements, the extreme values derived from LTS have a positive deviation of 0.2℃ and a lag of 2.5 minutes and 3.2 minutes for maximum temperature and minimum temperature, respectively. Differences of extreme values would be reduced with the enhancement of ambience wind speeds, even reduced to 0.1℃ when the speed is stronger than 4.5 m·s^-1. The deviation correction, which is developed mainly based on wind speed, reduces the difference to 0.03℃ and 0.01℃, and increases the consistent rate to 95.2% and 94.1% for the maximum temperature and minimum temperature, respectively.