We have recently demonstrated anomalously large electrostriction, a nonlinear induction of strain by an electric field, in three well-studied ionic conductors: Gd-doped ceria, (Nb, Y)-stabilized cubic bismuth oxide and Y-doped hydrated barium zirconate. At room temperature, all three materials exhibit a strain electrostriction coefficient (M11) exceeding 10-17 m2/V2 and can generate stress of tens of MPa without apparent signs of saturation. According to the theory of classical electrostriction, presented two decades ago by Prof. R. Newnham (Penn State), the electrostriction polarization coefficient (Q) scales with the ratio of elastic compliance to dielectric constant. This theory successfully describes most classes of materials from polymers, generating large strain and small stress, to relaxor ferroelectrics that generate small strain and large stress. However, the three ionic conductors mentioned above have a large elastic modulus (> 80 GPa) and moderately low dielectric constant (<100), which places their Q coefficients at least two orders of magnitude above the values predicted by the classical theory. Of these three materials, Gd-doped ceria has been the subject of the most detailed investigation, including in-situ high-resolution, room temperature XANES and EXAFS. According to these data, oxygen vacancies giving rise to elastic dipoles play a central role in the appearance of electrostriction. CeCe-oxygen vacancy (Vo) repulsion increases the CeCe-VO distance and forms six anomalously short CeCe-oxygen (OO) bonds. The resulting 7OO-CeCe-VO complex behaves as a strong elastic dipole with uniaxial symmetry. The components of the elastic tensor estimated from the differential EXAFS data are -4.7% and +11%. Application of an external electric field reduces the CeCe-VO repulsion, causing reversal of the local distortions. Because of the large strains involved, very large macroscopic stresses, which can reach hundreds of MPa, are generated, in spite of the fact that only a few percent of anion-cation complexes are involved. However, the presence of oxygen vacancies is not a sufficient condition for non-classical electrostriction. Oxygen deficient ceria (Ce(III)-doped) and Lu-doped ceria with the same concentration of vacancies as Gd-doped ceria exhibit much lower electrostriction, while Eu-doped ceria displays electrostriction similar to that of Gd doped ceria. Since Gd and Eu have almost identical ionic radii while Ce(III) is much larger and Lu(III) is much smaller, the radius of the dopant seems to be a critical parameter. Similar to Gd-doped ceria, (Nb, Y)-stabilized cubic Bismuth oxide has a fluorite structure and exhibits large non-classical electrostriction, which monotonically increases wihin the range of 16-23% of the oxygen vacancies. It is interesting to note that increase in the vacancy concentration from 16 to 23 % causes almost 50% decrease in the Young’s modulus, in the cubic Bismuth oxide, whereas in Gd-doped ceria increase in the vacancy concentration decreases the Young’s modulus by less than 10%. Similar differences between cubic Bismuth oxide and Gd-doped ceria are observed for Poisson’s ratio: it decreases two fold for the cubic Bismuth oxide and remains unchanged for Gd-doped ceria. These differences indicate that the influence of the large concentration of vacancies on electrostrictive and mechanical properties differ considerably from material to material. This is also supported by the fact that the electrostriction coefficient of 14.5 mol% Y-doped zirconia falls very close to the expected from the classical electrostriction. Hydrated Y-doped BaZrO3 is a well-studied protonic conductor in which OH groups can be present at »7% of the oxygen sites. Preliminary investigations indicate that this material also exhibits large non-classical electrostriction; yet hydrated Y-doped BaZrO3 at room temperature in fact does not contain any vacancies. On this basis, one can only conclude that a primary condition for large non-classical electrostriction is the presence of a large concentration of point defects, which can be of any type, but which must form elastic dipoles with a particular degree of freedom at room temperature. We suggest that a search for new electromechanically active materials must include those with a large concentration of such point defects in general and, ionic conductors, in particular.