South Africa's unemployment rate is one of the highest in the world, and it has important distributional implications. The paper examines both entry into and duration of unemployment using data for the mid-1990s. A probit model of unemployment shows an important role for race, education, age, gender, home-ownership, location, and numerous other variables, all of which have plausible explanations. The large race gap in unemployment is explored further by means of a decomposition analysis akin to that normally used to analyze wage discrimination. There remains a substantial residual which might represent unobserved characteristics, such as quality of education, or discrimination.