Under the scenario of climate change and rapid urbanization, the effects of urban green spaces on rainwater runoff have attracted special attention. The present studies highlighted the value of urban green spaces in reducing rainwater runoff but failed to consider the effects of landscape patterns (i.e., size, shape, or spatial arrangement of vegetation patches) on runoff reduction. The study mainly focuses on the landscape pattern changes in urban green spaces (including landscape types and metrics) as well as their effects on rainwater runoff detention. This paper investigates the spatial-temporal changes of urban green spaces within the 6th Ring Road of Beijing, and estimates their effects on rainwater runoff detention based on an empirical model. The result indicates that, the urban green space area decreased by 198.49 km~2 from 1040.67 km~2 in 2000 to 842.18 km~2 in 2010, which is mainly attributed to a sharp decrease of farmland and wasteland areas between the 5th and 6th Ring Roads. The largest patch index (LPI) and aggregation index (AI) of urban green spaces consistently decreased from 2000 to 2010, which implied the landscape patches become more isolated and fragmented. The volume of rainwater runoff controlled by urban green spaces slightly increased from 1.32 * 10~8 m~3 in 2000 to 1.46 * 10~8 m~3 in 2010 with the increase of summer rainfall, whereas the rainwater detention rate decreased from 23.43% to 16.89%. These changes are closely related to the decreases in green space area and landscape pattern metrics. In addition, the highest rainwater detention rate occurred in the urban green space between the 5th and 6th Ring Roads, however it decreased by 10% from 2000 to 2010. The urban green spaces between the 2nd and 3rd Ring Roads showed the lowest rainwater detention rate, and the flood risk should be paid highly attention. Therefore, an immense regional difference of rainwater runoff reduction is observed among five ring roads, which is closely related to the green space area and patch connectivity in different regions. The green space in the inner part of the 2nd Ring Road was composed primarily of grass and forest, and recorded large LPI and AI values, which promoted its role of rainwater runoff detention. By contrast, the green space between the 2nd and 3rd Ring Roads consisted of grass, forest, farmland, and wasteland, and the landscape metrics of the LPI and AI were relatively small, resulting in the poor capacity of rainwater runoff detention. Therefore, it is a good choice to optimize and adjust the composition and landscape pattern of urban green spaces. The results provide useful information for green space planning and management in highly urbanized areas. City managers should focus greatly on the role of urban green spaces in rainwater regulation and the scientific management of urban green spaces.