In wheat, a genome-wide association study (GWAS) combined with a study of genomic prediction (GP) was conducted for pre-harvest sprouting tolerance (PHST). For GWAS, a panel of 190 accessions was phenotyped for the three parameters of PHST (sprouting score, falling number and grain color) over two years and genotyped with 9,904 DArT-seq SNP markers. GWAS for main-effect quantitative trait nucleotides (M-QTNs) was performed using the following three models: CMLM, SUPER, and FarmCPU. A total of 171 M-QTNs (CMLM, 47; SUPER, 70; FarmCPU, 54) were detected, suggesting differences in efficiencies of the three models. As many as 15 epistatic QTNs (E-QTNs) involved in 20 first order epistatic interactions were also identified using PLINK v1.07. A total of 26 PHS-responsive genomic regions (each 5 Mbp in size) that spread over 16 chromosomes were demarcated, which overlapped the previously mapped QTLs, MTAs, cloned genes and the M-QTNs reported in the present study. As many as 20 definitive and stable QTNs were recommended for use in marker-assisted recurrent selection (MARS) for improvement of PHST. Candidate genes associated with different M-QTNs revealed key role of some QTNs in abscisic acid pathway involved in PHST. Genomic prediction accuracies (based on cross-validation approach) using three different models ranged from 0.41 to 055, which are comparable to the results of previous studies. In summary, the results of the present study improved our understanding of the genetic architecture of PHST in wheat and provided novel genomic resources for wheat breeding based on MARS and GP.