Estimates of the output gap are an important component of policy-makers’ toolkits. Both the theory underlying monetary policy analysis and the empirical models employed by central banks suggest that the output gap is a key variable explaining inflation. In this view, the estimate of the output gap provides not only an indication of how well the economy is operating relative to its potential, it also signals whether inflation is likely to increase or decrease in the future. The reliability of estimates of the output gap is therefore extremely important for policy making. However, a large literature has highlighted both conceptual and practical problems in measuring the gap, including the difficulty of using real-time data that will be revised in the future. The contribution of this paper is to assess the reliability of real-time estimates of the South African output gap, by estimating output gaps using a range of univariate methods applied to real-time gross domestic product (GDP) data. Consistent with the results of similar studies conducted in other countries, it is found that the real-time South African output gap estimates are in fact quite unreliable and are significantly revised over time. Furthermore, the source of these revisions is largely attributed to new data points becoming available, indicating the unreliability of end-of-sample estimates, rather than data or parameter revisions.