Recently, many breeders have preferred to use molecular markers for introgression backcross programs enabling foreground and background selection to cope with rapid cultivar changing of seed markets. In accumulation of target traits with marker-assisted selection, larger numbers of markers should give better resolution. For the analysis of quantitative traits, a high-density genetic map with a large number of markers is required for discovering more accurately linked markers with traits. Watermelon is a recalcitrant plant to generate a high-density genetic map with conventional molecular markers including simple sequence repeats (SSRs), since watermelon has narrow genetic diversity background and severe segregation distortions of those SSR markers. Thus, we have developed efficient and valid way to assemble genetic map and markers by next-generation sequencing coupled with genotyping by sequencing in F2 generation. After crosses between Citurullus lanatus ssp. citroides (PI254744 and PI189225) and C. lanatus ssp. lanatus (TS34, Korean cultigen), 163 of F2 progeny were sequenced through Illumina's Hi-Seq GAII platform. From sequence information of those variant call files, the SNPs were indexed and filtered by sequencing depth with genotype converter (SNP Genotyper), and optimized by heuristic physical bin mapping to construct more reliable genetic linkage map. Reliable SNP loci were determined and compared to sequences of physical reference map. Using the genetic map, we determined QTLs in F2:3 population and found major loci corresponding to seed size and powdery mildew race1 resistance in watermelon.