Objective: Body mass index is a complex genetic trait that involves multiple genetic loci. To identify genetic variants influencing body mass index in Korean population, we performed two staged combined mata-analysis of whole exome sequencing and exome array genotyping. Methods: We carried out whole exome sequencing in 917 subjects (619 type 2 diabetes [T2DM] and 298 non-diabetic) and identified 640,382 variants. Among these, we selected 148,371 variants for further genotyping using exome genotyping array. In the second stage, exome array genotyping was performed using Affymetrix Axiom® Biobank Plus Genotyping Array in 3,047 subjects (2,027 T2DM and 1,020 non-diabetic). Linear regression analysis for genetic association of body mass index was performed adjusting for age, sex, and T2DM status. Meta-analysis for stage 1 and 2 results was performed using METAL software. Results: In the combined meta-analysis for body mass index, none of the variants reached exome-wide significance of P < 2.0 × 10-6. However, four of non-synonymous genetic variants showed suggestive association with body mass index (chr15:41643250 G >T in NUSAP1, P = 3.04 × 10-5; rs75370284 in USP35, P = 4.57 × 10-5; rs1150781 in C6orf1, P = 5.48 × 10-5; rs559699688 in C20orf43, P = 6.01 × 10-5). Some of known loci (such as rs369602107 in TCF7L2, rs201228341 in RNF10, chr10:7605125 C > G in ITIH5 and rs78406786 in THNCL2) which were previously reported in other populations were associated with body mass index in this study with nominal significance (P < 0.05). Conclusion: This study validates previous reports of loci associated with body mass index and identified suggestive candidate regions involved body mass index in Korean population.