The study aims to investigate sleep duration in four different population groups in a national probability sample of older South Africans who participated in the Study of Global Ageing and Adult Health (SAGE) Wave 1. A national population-based cross-sectional study with a sample of 3284 aged 50 years or older in South Africa was conducted in 2008. The questionnaire included socio-demographic characteristics, health variables, and self-reported sleep duration. Results indicate that White Africans compared to other population groups had the lowest mean sleep duration (7.88 h among men and 7.46 h among women). The prevalence of short sleep was the highest among both men and women among the White African (18.8% in men and 16.9% in women) and Indian or Asian African population groups (14.5% in men and 17.1% in women), and lowest among both men and women in the Black African (7.0% in men and 6.5% in women) and multi-ancestry population groups (15.6% in men and 12.7% in women). The prevalence of long sleep was among both men and women the highest in the Black African population group (56.2% in men and 58.5% in women), and the lowest in the White African population group (36.4% in men and 24.3% in women). In a Poisson regression model, adjusted for sociodemographics and chronic disease status, coming from the male and female White African population group was associated with short sleep. In addition, coming from the Indian or Asian African population group was associated with short sleep. No population group differences were found regarding long sleep prevalence. White Africans reported more short sleep duration than the other population groups, while there were no racial or ethnic differences in long sleep. White Africans are more likely to have sleep durations that are associated with negative health outcomes. An explanation of the high short sleep prevalence among White Africans may be related to their racial or ethnic minority status in South Africa.