Situations often arise in which we are interested in measuring divergence between distributions over time or across regions. This paper proposes an entropy measure of divergence and derives analytical expressions for several parametric distributions widely used in modelling income distributions. An empirical application using income distribution data for the US from 1970 to 2019 shows a significant divergence or shift during the 1980-2000 period but relative stability on either side of this period.