Grain size and morphology is one of the main factors that determine the wheat yield. It is of great significance to mine the associated loci of grain size and morphology related traits and screen related candidate genes for improving wheat yield. In this study， 300 representative natural populations of winter wheat were used to identify the phenotype at nine grain traits， including 1000-grain weight， grain length， grain width， grain thickness， grain length-width ratio， grain area， grain perimeter， grain shape and grain plumpness. This collection was genotyped by 90K SNP chip， followed by genome-wide association study using mixed linear model （MLM+Q+K）. The results showed that these traits showed rich phenotypic variation， with the coefficient of variation ranged from 3.80%~26.06% and the broad heritability from 56.25% to 91.98%. Sixty-six loci （P≤0.001） were detected by GWAS on 18 chromosomes except 3D， 4D and 5D， explaining 3.74%~14.34% of phenotypic variation at each locus. Thirty-seven pleiotropic loci were found associating with two or more grain traits. Especially， the BS00022512_51 marker on chromosome 3B was simultaneously correlated with 4 grain traits （grain length， grain width， grain thickness and grain length-width ratio）?， with the largest apparent contribution rate （7.06%~14.34%）. The wsnp_Ex_c4480_8055475 markers on chromosome 6D were simultaneously associated with six grain traits except grain thickness， grain shape and grain plumpness， and the phenotypic contribution rate was 3.81%~8.25%. Haplotype analysis of BS00022512_51 and wsnp_Ex_c4480_8055475 markers showed that there were three haplotypes of GC-Hap1， AT-Hap2 and AC-Hap3 in wsnp_Ex_c4480_8055475 loucs on chromosome 6D， and Haplotype GC-Hap1 is a haplotype with large grains and high 1000-grain weight. The three haplotypes accounted for 65.58% and 32.25% and 2.17% in population， respectively. Haplotype GC-Hap1 was predominant in four winter wheat regions in China. Nine candidate genes were proposed based on 37 significantly-associated multiple effect SNPs.