Abstract
The melting temperatures of IV-IV, III-V, and some II-VI crystals are predicted with an accuracy generally better than 10%. Only empirical data for Si are used to fix scaling constants.