Summary Until recently, achieving a reference‐quality genome sequence for bread wheat was long thought beyond the limits of genome sequencing and assembly technology, primarily due to the large genome size and > 80% repetitive sequence content. The release of the chromosome scale 14.5‐Gb IWGSC RefSeq v1.0 genome sequence of bread wheat cv. Chinese Spring (CS) was, therefore, a milestone. Here, we used a direct label and stain (DLS) optical map of the CS genome together with a prior nick, label, repair and stain (NLRS) optical map, and sequence contigs assembled with Pacific Biosciences long reads, to refine the v1.0 assembly. Inconsistencies between the sequence and maps were reconciled and gaps were closed. Gap filling and anchoring of 279 unplaced scaffolds increased the total length of pseudomolecules by 168 Mb (excluding Ns). Positions and orientations were corrected for 233 and 354 scaffolds, respectively, representing 10% of the genome sequence. The accuracy of the remaining 90% of the assembly was validated. As a result of the increased contiguity, the numbers of transposable elements (TEs) and intact TEs have increased in IWGSC RefSeq v2.1 compared with v1.0. In total, 98% of the gene models identified in v1.0 were mapped onto this new assembly through development of a dedicated approach implemented in the MAGAAT pipeline. The numbers of high‐confidence genes on pseudomolecules have increased from 105 319 to 105 534. The reconciled assembly enhances the utility of the sequence for genetic mapping, comparative genomics, gene annotation and isolation, and more general studies on the biology of wheat. Significance Statement This new release of bread wheat cv. Chinese Spring reference genome sequence, IWGSC RefSeq v2.1, features correction of assembly errors affecting approximately 10% of the prior IWGSC RefSeq v1.0 release using genome‐wide optical maps and filling of gaps with single‐molecule long‐reads as well as incorporating re‐annotation of TEs and re‐computation of gene coordinates. These refinements enhance the sequence utility for breeding and research applications.