Overexpression of rab5 via a Sindbis virus vector resulted in a 2-3-fold stimulation of horseradish peroxidase uptake in BHK-21 cells. Based on this functional assay of rab5 activity, we conducted extensive mutational analysis of the structure-function relationship of rab5. A total of 21 deletion and substitution mutations were created and their effects on rab5 activity were examined. Deletion of the entire C-terminal tetrapeptide motif CCSN abolished rab5 activity. A mutant with the last three residues deleted, however, showed residual rab5 activity. Truncation of only two residues from the C terminus had no effect on the biological activity of rab5. A mutant containing a 4-residue deletion from the N terminus retained full activity in comparison with wild-type rab5. N-terminal deletion of 19 residues only partially blocked rab5 activity. Substitution mutations in the guanine nucleotide binding motifs showed dramatic effects on rab5 function. In addition to the previously reported N133I mutation, the S34N mutation also resulted in a guanine nucleotide binding defective form that was a dominant inhibitor of endogenous rab5 activity. The Q79L mutation (the ras equivalent Q61L decreases intrinsic and GTPase-activating protein-stimulated GTPase activities), however, had no effect on rab5 activity. The S35N mutation, which is immediately downstream of the first GTP/GDP binding motif, decreased guanine nucleotide binding by approximately 4-fold and partially inactivated rab5. Mutations in several other conserved residues (K22A, F57Y, and R81A) also resulted in partial loss of rab5 activity. Eight mutations in and around the putative effector domain had little effect on rab5 activity. In light of these data, the structure-function relationship of rab5 is discussed and compared with that of ras, the prototype of small GTPases.