Kernel ratio (KR) is closely associated with grain yield per ear in maize (Zea mays), and the analysis of genetic mechanism for kernel ratio is important for high yield breeding. We used 309 inbred lines of maize as an association population, and used FarmCPU (fixed and random model circulating probability unification), CMLM (compressed mixed linear model), and MLMM (multiple loci mixed model) to conduct genome-wide association studies for kernel ratio of maize grown in Yuanyang, Dancheng, and Sanya in 2017 and 2019, and best linear unbiased estimate (BLUE). Eighteen significant SNPs for KR were identified (P < 1.72E-05), which were located on Bin1.02(1), Bin1.08(1), Bin1.12(1), Bin2.05(5), Bin2.06(1), Bin2.10(1), Bin4.08(1), Bin4.09(1), Bin5.02(1), Bin5.04(1), Bin5.05(1), Bin6.05(1), Bin8.03(1), and Bin9.04(1). The phenotypic variation explanation for KR of the 18 SNPs ranged from 0.067% to 15.43%. Five SNPs (S1_304584425, S5_11751831, S5_93814060, S5_186385476, and S8_94354503) varied from 10.09% to 15.43%, and were considered as major SNPs. S2_87292896 was detected by CMLM and MLMM in BLUE environment and Yuanyang 2019. S2_111319193 was identified by FarmCPU and MLMM in BLUE environment. S5_93814060 was detected by CMLM and MLMM in Dancheng 2017. Bin1.08, Bin2.06, Bin4.09, and Bin6.05 might be important genomic regions for KR compared with results of previous studies. A total of 32 candidate genes were identified, among which E3 ubiquitin-protein ligase UPL1, DEAD-box ATP-dependent RNA helicase 52 RH52, protein kinase homolog4, SNARE-interacting protein KEULE, and elongation factor (EF1A) may be important genes for KR.