The existing theories of the resistivity of mixtures assume regular arrangements of the two components, rather than random mixtures. A theory for a random mixture is given, based on the assumption that each crystal acts as if surrounded by a homogeneous medium whose properties are those of the mixture. Comparisons with experiment are made. The experimental data that have been examined fall roughly into two classes. One class consists of mixtures, where the variation of resistivity with composition disagrees violently with this theory, making it clear that the assumptions made are completely inapplicable. The remaining class consists of mixtures which generally agree well with the theory.