Wheat (Triticum aestivum L.) is one of the most important food crops in the world. The kernel number per spike of wheat can directly affect yield. F6 lines derived from the cross of the female Avocet and the male Chilero were used to detect QTLs for kernel number per spike (KNS) in five different environments. In this study, the phenotypic data of kernel number per spike (KNS) based on five different environments of field trials showed a wide variation and approximated a normal distribution. With a 55K single nucleotide polymorphism (SNP) array resulting in 1,029 SNP markers that were used to construct genetic map of 3,859.32 cM with the average distance of 3.75 cM between markers. A total of 6 QTLs were identified for kernel number per spike trait. The QTLs mapped to chromosomes 3B, 5A, 6A, 7B, 7D explained 7.76%-19.20% of the phenotypic variation in single environment. With consistent QTL (that are identified in multiple environments and Blue) on chromosome 6A in RIL population. The consistent QTL explained 8.88%, 11.18%, 11.42%, 11.61% and 19.20% of the phenotypic variances for KNS, respectively. Combining annotation information of the IWGSC RefSeq, a total of 138 high confidence genes with a physical length of 12.6 Mb were identified. Among these genes, TraesCS6A02G092400 and TraesCS6A02G092800 were finally associated with kernel number per spike, which encoded ripening-related protein. The single-nucleotide polymorphism (SNP) markers AX-110947207, AX-111006064 and AX-108982274 were closely linked to the consistent QTL and could be used for marker-assisted selection (MAS) for grain yield in wheat breeding programs. These results provide direct genetic evidence that QTLs and candidate genes can contribute to further understanding of target traits, fine mapping and developing molecular markers.