Interviewers administered the Rosenberg Self-Esteem scale (RSES) to five groups of Black (formal township and informal settlement), White, Indian, and mixed race adult residents of Greater Pretoria. The results demonstrated that the RSES was psychometrically sound for the five groups. The minimal effects of sociodemographic characteristics on global self-esteem showed that the RSES and its two dimensions, self-competence (SC) and self-liking (SL), were suitable in this setting. All five groups scored above the theoretical midpoint of the RSES, indicating that generally positive self-evaluations appear to be universal. The relationships between positively and negatively worded items, SC, and SL attested to the following: internal structure reliability, congruence between positive and negative items, no negative biases in response, and concordance between SC and SL dimensions. The significant differences between informal settlement residents and the other four groups on global self-esteem, positively and negatively worded items, and SC and SL were possibly due to physiological needs taking precedence over higher order needs.