The combination of cis-protected metal fragments with linear linkers is expected to yield molecular squares. We found instead that treatment of the 900 angular precursor trans-[RUCl2(dmso-S)(4)] (1) with an equivalent amount of the linear and rigid pyrazine (pyz) linker unexpectedly yields, in a number of different experimental conditions, the molecular triangle [{trans, cis-RUCl2(dmso-S)(2)(mu-pyz)}(3)] (3), together with polymeric material. Very similar results were also obtained from the reaction between 1 and the preformed corner fragment trans, cis, cis-[RUCl2(dmso-S)(2)(pyz)(2)] (6). In both cases, the expected molecular square [{trans, cis-RuCl2(dmso-S)(2)(mu-pyz)}(4)] (4) was observed only as a transient species. These results suggest that 3, which is the first example of a neutral molecular triangle with octahedral metal corners and pyrazine edges, is both the thermodynamic and the kinetic product of the reactions described above. The X-ray structure of 3 shows that the main distortions from ideal coordination geometry concern the N-Ru-N angles, which are narrower than 900, and the coordination bonds of pyz. The pyrazine molecules, which are basically planar, are significantly tilted from linearity. Calculations performed on 6 indicated that the N-Ru-N angle is ca. six times more rigid than the tilt angle of pyrazine. The structural and theoretical findings on 3 and 6, together with the previous examples of molecular triangles and squares with cis-protected metal corners and linear pyz edges, suggest that the entropically favored molecular triangles might be preferred over the expected molecular squares with metal corner fragments that spontaneously favor N-pyz-M-N-pyz angles narrower than 900 because of the presence of ancillary ligands with significant steric demand on the coordination plane. The rather-flexible coordination geometry of pyrazine can accommodate the moderate distortions from linearity required to close the small metallacycle with modest additional strain.