Abstract: The spatial pattern of landscape is fundamental to landscape's function and development. Landscape metric analysis is widely applied in fields such as land use/land cover patterns, ecosystem service, forest monitoring, urban sprawl, regional biodiversity, etc. There are high correlations between many presented landscape metrics, which not only result in redundancy but also influence the clarity and accuracy of interpretation. Meanwhile, most of previous researches on landscape metric analysis use raster data and few use vector data. In order to solve these problems, a case study of Guangzhou land use pattern was performed. Firstly, we calculate 44 common landscape metrics using vector data. Then, multivariate statistical analysis, a combination of Spearman correlation analysis and factor analysis, is applied to solve the problem of redundancy. A core set of six landscape metrics in vector format is extracted, which respectively describe landscape fragmentation, average patch size, average shape complexity, spatial isolation, shape complexity variation and land use similarity.