Genetic analyzes and utilization of wild genetic variation in common bean (Phaseolus vulgaris L.) for crop improvement have been hampered by evaluation difficulties, identification of advantageous variation, and linkage drag. The lack of adaptation to field conditions and the existence of highly structured populations make association mapping of diversity panels not optimal. Joint linkage mapping of nested populations avoids the later constraint, while populations crossed with a common domesticated parent allow the evaluation of wild variation within a more adapted background. We developed three domesticated by wild backcrossed inbred line populations (BC₁S₄), using three wild accessions representing the extreme range of rainfall of the Mesoamerican wild bean distribution crossed to the elite drought tolerant domesticated parent SEA 5. We evaluated the populations under field conditions in three environments, two fully irrigated trials in two seasons and a simulated terminal drought in the second season. The goal was to test if these populations responded differently to drought stress and to detect yield-associated genomic regions. Our results revealed that the populations from the wild parents of the low rainfall part of the distribution showed higher yield. We found 20 QTLs for yield in 13 unique regions on eight of the 11 chromosomes of common bean. Five of these regions showed at least one wild allele that increased yield over the domesticated parent. The variation explained by these QTLs ranged from 0.6 to 5.4 % of the total variation and the additive effects ranged from -164 to 277 kg ha^-1, with evidence suggesting allelic series for some QTLs. The average allele effects from the parent of the wettest environment were lower through all the test environments. Our results underscore the potential of wild variation for bean crop improvement as well the identification of regions for efficient marker-assisted introgression.