Quantitative studies of mental health in South Africa typically find that Black African adults report significantly more depressive symptoms than other adults on average. However, insufficient attention has been paid to differences among Black African adults, and to how the distribution of depressive symptoms compares between these adults and other adults. We augment existing research by investigating how the size and composition of the race gap in depression scores varies at different percentiles of the mental health distribution. We analyze national self-reported data on the frequency of depressive symptoms, and estimate recentred influence function decompositions of the unconditional race gap at each quantile of the distribution. The analysis identifies a race gap in depression scores that is twice as large at the lower tail of the mental health distribution than at the upper tail, signalling that Black African adults are particularly less likely than other adults to report no, or only a few, symptoms. At the lower tail, the race gap derives mostly from differences in the relationship between characteristics and mental health, while at higher percentiles, differences in the level of endowments, or stressors, are more important. At the upper tail, where depression scores are likely to correspond to major depressive episodes, the race gap narrows to zero. The size and composition of the average race gap in depression scores therefore masks considerable variation about the mean, which should be considered when policy is targeted to redress inequalities in mental health treatment.