Drought is one of the major abiotic constraints seriously influencing bread wheat (Triticum aestivum L.) genotype production in Ethiopia. Genetic resources and selection methodologies are among the prerequisites to improve the efficiency of breeding for drought tolerance. The objectives of this study were to determine the principal selection indices for drought and to identify drought tolerant genotypes under drought conditions. 256 bread wheat genotypes were evaluated using a simple lattice design with two replications. Number of irrigation and the amount of water supply was similar for both water regimes until 50% heading stages. Non-stressed plots were irrigated 2 times after 50% heading stage, while stressed plots received no water in order to simulate terminal drought. Genotypes showed highly significant differences (P ≤ 0.01) for grain yield under non-stressed and stressed conditions. Principal component and correlation analyses revealed mean productivity, geometric mean productivity, grain yield index and stress tolerance index as the principal indices highly correlated with grain yield in the stressed and non-stressed environments, indicating their suitability for identifying superior genotypes. Genotypes 147 and 100 were identified as more tolerant, which could be useful for drought stress tolerance breeding. Cluster analysis classified the genotypes into nine clusters. Cluster IX consisted of four genotypes, 18,137,100 and 147 gave high grain yield under both the moisture -stressed and non-stressed conditions with high value of mean productivity, geometric mean productivity, grain yield index and stress tolerance index. Therefore, breeders can select suitable genotypes under water-stressed conditions and compare their performance under non-stressed conditions using MP, GMP, YI and STI indices as a means to combine information on performance under both conditions.
