Abstract: The human face develops from neuroectoderm, and it is one of the most significant characteristics of individual recognition. The study of facial features and their genetic factors could provide help for biomedicine and forensics etc. Although some genes have been identified for various normal and abnormal facial phenotypes, there are still a lot of genes needing to be discovered. In this study, a GWAS using about 700 000 genome-wide markers from 1 140 Peruvians in CANDELA（Consortium for the analysis of the diversity and lvolution of Latin America）was conducted. Many traits were assessed in these samples using 47 landmarks to obtain quantitative measurements, such as middle and lower facial areas, lateral Euclidean distances between 47 landmarks. The results were also tested in a different 384 Brazil ＆ Chile CANDELA samples. Association tests were conducted by multivariate linear regression with an additive genetic model considering age, sex, BMI（Body mass index）, and five genetic PCs（Principal component） as covariates. The result shows that 17 SNPs（Single nucleotide polymorphisms） situated in LD gene regions（EDAR, LIMS1, RANBP2, SULT1 C4, CCDC138, GCC2） show strong associations with 7 quantitative phenotypes. EDAR is a pleiotropic gene affecting the development of head and face, which affects the middle craniofacial morphology in this study.