Abstract Two new families of matrix variate distributions are introduced. They are based on matrix normal distribution and yet can be used to model data involving skewness. The properties of the two families are investigated. Among others, the marginals, conditionals, stochastic representation, linear and quadratic forms are studied.