The DEAD-box family of putative RNA helicases is composed of ubiquitous proteins that are found in nearly all organisms and that are involved in virtually all processes involving RNA. They are characterized by two tandemly linked, RecA-like domains that contain 11 conserved motifs and highly variable amino- and carboxy-terminal flanking sequences. For this reason, they are often considered to be modular multi-domain proteins. We tested this by making extensive BLASTs and sequence alignments to elucidate the minimal functional unit in nature. We then used this information to construct chimeras and deletions of six essential yeast proteins that were assayed in vivo. We purified many of the different constructs and characterized their biochemical properties in vitro. We found that sequence elements can only be switched between closely related proteins and that the carboxy-terminal sequences are important for high ATPase and strand displacement activities and for high RNA binding affinity. The amino-terminal elements were often toxic when overexpressed in vivo, and they may play regulatory roles. Both the amino and the carboxyl regions have a high frequency of sequences that are predicted to be intrinsically disordered, indicating that the flanking regions do not form distinct modular domains but probably assume an ordered structure with ligand binding. Finally, the minimal functional unit of the DEAD-box core starts two amino acids before the isolated phenylalanine of the Q motif and extends to about 35 residues beyond motif VI. These experiments provide evidence for how a highly conserved structural domain can be adapted to different cellular needs.