Multiple regression and stepwise regression methods have been used to study the relations beween Beijing urban temperature and the structure of urban surface during daytime and nighttime in different seasons. The data used in this study consisted of day and night temperatures observed on 30 observation sites during spring, summer, autumn and winter of 1982, and the coverage of green areas, building areas and water bodies in the 1000m?1000m area around each observation site. It was found out that Beijing urban temperature had significant positive correlation with the coverage of building areas and negative correlation with the coverage of green areas, but the degree of correlation varies from day to night and from season to season. It was shown that as the coverage of green areas increased, the decrease of temperature was most obvious in summer daytime, and as the coverage of building areas increased, the increase of temperature Was most obvious in winter nighttime. It was also shown that the correlation between temperature and water body was very poor mainly due to the lack of large water body near the observation sites, which means that small water body in urban areas has no significant influence on urban temperature.