Starch is the most important component in endosperm of sorghum grain.  Usually, two types of starch are present: amylose (AM) and amylopectin (AP).  The levels of AM and AP contents play a significant role in the appearance, structure, and quality of sorghum grains and in marketing applications.  In the present study, a panel of 634 sorghum (Sorghum bicolor (L.) Moench) accessions were evaluated for starch, AM, and AP contents of grain, which included a mini core collection of 242 accessions from the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) in India, and 252 landraces and 140 cultivars from China.  The average starch content was 67.64% and the average AM and AP contents were 20.19 and 79.81%, respectively.  We developed a total of 260 000 high-confidence single nucleotide polymorphism (SNP) markers in the panel of 634 accessions of S. bicolor using specific locus amplified fragment sequencing (SLAF-seq).  We performed genome-wide association studies (GWAS) of starch, AM, and AM/AP of grain and SNP markers based on a mixed linear model (MLM).  In total, 70 significant association signals were detected for starch, AM, and AM/AP ratio of grain with P<4.452×10^–7, of which 10 SNPs were identified with significant starch, 51 SNPs were associated with AM, and nine SNPs were associated with the AM/AP ratio.  The Gene Ontology (GO) analysis identified 12 candidate genes at five QTLs associated with starch metabolism within the 200-kb intervals, located on chromosomes 1, 5, 6, and 9.  Of these genes, Sobic.006G036500.1 encodes peptidyl-prolyl cis-trans-isomerase CYP38 responsible for hexose monophosphate shunt (HMS) and Sobic.009G071800 encodes 6-phospho-fructokinase (PFK), which is involved in the embden-meyerhof pathway (EMP).  Kompetitive allele specific PCR (KASP) markers were developed to validate the GWAS results.  The C allele is correlated with a high starch content, while the T allele is linked with a low level of starch content, and provides reliable haplotypes for MAS in sorghum quality improvement.