Urban Heat Island (UHI) is one of the major problems in the 21^st century posing to human beings as a result of urbanization and industrialization. It is primarily triggered by the dense built environment (i.e., replacement of the natural landscape) as well as anthropogenic heat in cities, and has led to huge negative impacts on human life. Urban parks, which are known as "Urban Cool-island", have been considered as an effective measure in alleviating UHI effects and improving urban thermal environment. Nevertheless, as urban land use is of high tension, it has become a serious issue during the planning and designing process of city parks that how to maximize the ecological functions of the park landscape, and to make it effective in improving the urban ecology and urban climate. In this study, 24 city parks in the urban area of Beijing were selected to explore the effects of spatial characteristics of city parks (including the landscape composition, the patch morphology and the spatial distribution) on internal and external thermal environment, including the temperature inside the parks (T_a), sphere of influence on the surrounding environment (L_max) and the maximum cooling range (ΔT_max). Temperature distribution was inversed from Landsat-5 TM remote sensing data. The results showed that from the perspective of the landscape composition, T_a, L_max and ΔT_max were positively correlated with water areas at a significant level, which was considered as the key factor affecting the thermal environment both inside and outside the parks. T_a and ΔT_max represented no significant correlation with the forest and lawn areas. Comparatively, they were closely related with the greenness and the proportion of impermeable surface. Conversely, L_max had no obvious correlation with the greenness, but was substantially positively correlated with the forest area. Hence, in order to maximize the cooling effects both inside and outside the parks, it would be of necessity to enrich the inner spatial structure of green space and increase the three-dimensional greenness to the greatest extent under the premise of ensuring a certain size of the green space area. With respect of the patch morphology, the more complex the geometric shape of green patch was, the lower the internal temperature and the further the sphere of influence would be. The boundary shape of the park displayed a certain correlation with the internal temperature, but had no distinct impacts on the surrounding thermal environment. In terms of the spatial distribution, T_a, L_max, and ΔT_max were all remarkably associated with the distribution of impermeable surface. That is, the more dispersed the impermeable surface arrangement was, the lower the internal temperature and the larger sphere of influence and cooling range would be. In addition, the parks with more dispersed forest distribution would bring about the lower internal temperature and the larger sphere of influence. While the forest distribution of the parks had no apparent impacts on ΔT_max. In conclusion, from the perspective of the urban heat island mitigation, the spatial landscape characteristics of parks should be taken as important considerations in the city park planning and design.