Identification of agronomically important genes is a critical step for crop breeding through biotechnology. Multiple approaches have been employed to identify new gene targets, including comprehensive screening platforms for gene discovery such as the over-expression of libraries of cDNA clones. In this study, random genomic fragments from plants were introduced into rice and screened for drought tolerance in a high-throughput manner with the aim of finding novel genetic elements not exclusively limited to coding sequences. To illustrate the power of this approach, genomic libraries were constructed from four grass species, and screening a total of 50,825 transgenic rice lines for drought tolerance resulted in the identification of 12 reproducibly efficacious fragments. Of the twelve, two were from the mitochondrial genome of signal grass and ten were from the nuclear genome of buffalo grass. Subsequent sequencing and analyses revealed that the ten fragments from buffalo grass carried a similar genetic element with no significant homology to any previously characterized gene. The deduced protein sequence was rich in acidic amino acid residues in the C-terminal half, and two of the glutamic acid residues in the C-terminal half were shown to play an important role in drought tolerance. The results demonstrate that an open-ended screening approach using random genomic fragments could discover trait genes distinct from gene discovery based on known pathways or biased toward coding sequence over-expression.