Commercial legume varieties vary in terms of their drought tolerance when grown as sole crops, though relatively little is known about how legume variety selection affects cereal–legume intercrop performance under drought conditions. This study aims to test the hypothesis that positive rhizosphere interactions in faba bean–wheat intercrops will confer a “buffering capacity” on faba bean and wheat performance under water stress and that this effect will (i) depend on faba bean varietal selection and (ii) be enhanced with increasing faba bean varietal diversity. In a greenhouse experiment, three commercial faba bean (Vicia faba L.) varieties [Gloria (G), Alexia (A), Julia (J)] were grown in sole crop or intercropped with spring wheat (Triticum aestivum L.) under well-watered or water-stress conditions. Under intercropping, either one, two, or all three faba bean varieties were grown together with wheat to test the effect of intraspecific diversity on a cereal–legume intercrop performance. Consistent with the proposed hypothesis, we found that, under well-watered and water-stress conditions, wheat and faba bean shoot biomass production and nitrogen (N) acquisition improved with intercropping and that faba bean variety and variety mixture strongly modulated the intercropping effect. Interestingly, in both well-watered and water-stress conditions, wheat dry biomass and N accumulation were greatest in intercrops containing Gloria, while nodule number, nodule weight, and N accumulation in faba bean were greatest for intercrops containing Alexia and Julia (AJ). The effect of varietal diversity was inconsistent. Intercrops with two faba bean varieties tended to have positive or neutral effects on measured wheat and faba bean variables. However, overall performance under intercropping was generally reduced when all three faba bean varieties were planted with wheat. The effect of faba bean species diversity can buffer faba bean–wheat intercrop performance against water stress, and intercropping tended to have positive or neutral effects on the measured wheat and faba bean variables, notably with two-varietal faba bean mixtures.