Accurate determinations of masses and radii in binary stars, along with estimates of the effective temperatures, metallicities, and other properties, have long been used to test models of stellar evolution. As might be expected, observational constraints are plentiful for main-sequence stars, although some problems with theory remain even in this regime. Models in other areas of the H-R diagram are considerably less well constrained, or not constrained at all. I summarize the status of the field, and provide examples of how accurate measurements can supply stringent tests of stellar theory, including aspects such as the treatment of convection. I call attention to the apparent failure of current models to match the properties of stars with masses of 1.1–1.7 M⊙ that are near the point of central hydrogen exhaustion, possibly connected with the simplified treatment of convective core overshooting.