This review is concerned with the structures and mechanisms of a superfamily of regulatory GTP hydrolases (G proteins). G proteins include Ras and its close homologs, translation elongation factors, and heterotrimeric G proteins. These proteins share a common structural core, exemplified by that of p21ras (Ras), and significant sequence identity, suggesting a common evolutionary origin. Three-dimensional structures of members of the G protein superfamily are considered in light of other biochemical findings about the function of these proteins. Relationships among G protein structures are discussed, and factors contributing to their low intrinsic rate of GTP hydrolysis are considered. Comparison of GTP- and GDP-bound conformations of G proteins reveals how specific contacts between the gamma-phosphate of GTP and the switch II region stabilize potential effector-binding sites and how GTP hydrolysis results in collapse (or reordering) of these surfaces. A GTPase-activating protein probably binds to and stabilizes the conformation of its cognate G protein that recognizes the transition state for hydrolysis, and may insert a catalytic residue into the G protein active site. Inhibitors of nucleotide release, such as the beta gamma subunit of a heterotrimeric G protein, bind selectively to and stabilize the GDP-bound state. Release factors, such as the translation elongation factor, Ts, also recognize the switch regions and destabilize the Mg(2+)-binding site, thereby promoting GDP release. G protein-coupled receptors are expected to operate by a somewhat different mechanism, given that the GDP-bound form of many G protein alpha subunits does not contain bound Mg2+.