This paper re-evaluates the recent criticisms of ‘Thirlwall's law’ against the literature on growth transitions. The unpredictable nature of growth transitions in developing economies suggests that the evidence derived from single-regime regression models, on which critics have based most of their arguments, is only suggestive about the long-run causes of growth. A rigorous test of Thirlwall's law requires a more in-depth analysis of turning points in a developing country's growth performance, and whether the growth law accurately predicts the sustainability of growth transitions. These arguments are illustrated with an application to South Africa over the period 1960–2017. The results show that it is misleading to evaluate Thirlwall's law across a single regime. Once regime shifts are controlled for, the growth law accurately predicts South Africa's growth performance during 1977–2003, and sheds light on the sustainable and unsustainable nature of growth transitions across the sub-periods 1960–1976 and 2004–2017, respectively. Since the literature on growth transitions identifies a competitive exchange rate as an initiating source of growth, rather than an individual long-run determinant, the omission of the level of the real exchange rate from the original growth law should not be regarded as a major weakness.