Recent high dropout and low graduation rates in the South African higher education institutions as well as government funding cuts and the economic uncertainty due to COVID-19 pandemic have heightened the urgency for the higher education sector to improve its productivity. However, empirical evidence on the productivity growth of the sector remains unexplored. To address this gap, we applied a Färe-Primont index approach to a panel data of 22 public universities over an 8-year period to measure total factor productivity (TFP) and its components--technological change, technical, scale and mix efficiency changes. We also used a feasible generalised least squares model to assess the determinants of productivity and efficiency growth. The results show that the average TFP of the sector for the study period was 0.631, led by historically advantaged universities (0.894), whilst historically disadvantaged universities had lower average TFP (0.823). During the period, TFP increased by 3.43%, largely driven by scale and mix efficiency changes (5.32%) and technical efficiency change (0.83%), whilst technical change declined by 1.80%. In terms of university types, the comprehensive universities achieved the largest TFP growth (6.13%) followed by traditional universities (4.85%), and technology universities by 1.41%. TFP growth was positively influenced by student graduation rates, quality of academics and academic-student ratios. Therefore, policy considerations to improve the sector's productivity and efficiency should consider investment on research and development, adoption of teaching and research innovations, re-skilling through training and education and aligning admission policies with staffing.