South Africa exhibits one of the highest levels of income inequality globally, reflecting persistent spatial exclusion. This paper examines the extent and causes of spatial inequality using household microdata, microsimulations, and a structural spatial general-equilibrium model. The microdata analysis indicates that inequality within (rather than across) urban and rural areas accounts for the majority of overall inequality, with long commuting times strongly associated with higher unemployment and lower incomes. The simulation analysis suggests that reducing commuting costs can meaningfully lower unemployment and inequality. The structural modeling results point to transportation and housing policies as effective tools to promote spatial integration and reduce income inequality.