Phenotyping maize (Zea mays L.) for drought tolerance is costly and time consuming. Our objectives were to determine (i) the heritability, genetic variance, and genetic correlations for grain yield and secondary traits in maize under drought and (ii) the efficiency of indirect selection through secondary traits versus genomewide selection. Testcrosses of 238 recombinant inbreds from the intermated B73 × Mo17 population were evaluated in multilocation trials under managed drought and nondrought (control) conditions in Minnesota in 2009 and 2010. Mean grain yield under drought was 52% of the mean grain yield in the control experiments. Heritability for grain yield was 0.37 ± 0.08 under drought and 0.60 ± 0.04 in the control experiments. Indirect selection based on anthesis-silking interval, leaf senescence, leaf chlorophyll content, or grain yield in control conditions was not predicted to be more efficient than direct selection for grain yield under drought. Genomewide selection (with 998 markers) for grain yield under drought had a predicted relative efficiency of 1.24. Genetic correlations estimated from genomewide marker effects agreed well with correlations estimated from genetic covariances. Given that multiple cycles of marker-based selection can be done per year in maize and that genotyping is cheaper than phenotyping for drought tolerance, our results suggest that genomewide selection could increase genetic gains per unit time for grain yield under drought.