Drought stress is one of the major factors limiting yield and yield stability of durum wheat. Classical breeding schemes have used direct selection for yield over multiple locations or drought-related proxy traits, as physiological ones, but low heritability and high genotype x environment interaction have greatly limited the effectiveness of these approaches for durum wheat. During last years, genomics and other omic approaches have provided a considerable contribution to identify markers, transcripts, peptides, and metabolites associated with responses to drought. As whole genome sequences, physical maps, genetic and functional genomic tools are now available also for tetraploid wheats, integrated approaches leveraging molecular breeding, genetic engineering, and gene editing can be fully exploited to improve yield in drought-prone conditions. On one hand, the identification of quantitative trait loci (QTLs) with a major effect and the cloning of the underlying loci (coding and non-coding sequences) are expected to move the selection toward a “breeding by design” approach that will accumulate an increasing number of known useful alleles into elite genotypes. Additionally, genomic selection approaches are promising as they leverage genome-wide loci also with very small effects and, as such, with no previous knowledge on their map position and function. Efficient integration of all the genomic tools and approaches in breeding programs will ultimately improve durum wheat to better sustain yield and quality in drought-prone environments while reducing the gap between yield potential and actual yield.