Quality of sexually transmitted disease (STD) primary care in South Africa varies widely but reasons for this are poorly understood. We investigated 37 randomly sampled clinics providing STD care, with simulated patients, and staff interviews and record review. Census data provided local socioeconomic indicators. Multiple regression identified independent predictors of quality. Of 271 simulated patient visits, 79% were correctly treated and 39% were correctly managed. Women received worse care, and care tended to be poorer in mainly African and mainly coloured (mixed race) areas. African and Indian nurses were more likely to provide correct treatment. Previous STD training was marginally associated with correct treatment. Quality assessments using simulated patients were not generally associated with assessments using staff interviews and record review. There were frequent missed opportunities for STD prevention and treatment, and evidence of racial but not socioeconomic inequalities.