South Africa is a water scarce country with deteriorating water resources. Faced with tight fiscal and water resource constraints, water utilities would have to adopt technically efficient water management technologies to meet developmental socio-economic objectives of universal coverage, aligned to the United Nation's Sustainable Development Goal 6. It is important to measure the technical efficiency of utilities as accurately as possible in order to inform policy. We do this by using a non-parametric method known as Data Envelopment Analysis to determine, measure, analyse and benchmark the technical efficiency of all water boards in South Africa. Our contribution to the literature is twofold: This is the first paper to model technical efficiency of water boards as utility suppliers and guardians of water services in South Africa, and second, we address the over- and underestimation issues of technical efficiency measurement in the water sector. We do this by modelling one of the most pronounced negative externalities from water provision (water losses) as an undesirable output using the approach developed by You & Yan. We find, on average, technical efficiency of water boards is 49%, with only three of the nine water boards technically efficient. Six of the smaller water boards showed high levels of inefficiency with an inefficiency rate of 51%, which is equivalent to wastage in expenditure of R3.7 billion. Six water boards operate at increasing returns to scale and two are scale efficient. Only Rand and Sedibeng water boards exhibited decreasing returns to scale. Therefore, redirecting potential efficiency savings to optimal uses could result in technical and scale efficiency for the sector. Scale efficiency results seem to support larger regional water boards as small- to medium-sized water boards are scale inefficient with low technical efficiency. For example, Amatola Water (small water board) with an efficiency score of only 16% has a total expenditure of 18% of that of Umgeni (large water board), but sells only 6.7% of the quantity sold by Umgeni. Amatola also has seven times the proportion of water losses compared with Umgeni and charges 1.6 times the tariff of Umgeni. The ratio model with an undesirable output outperforms previous methods to deal with undesirable (bad) outputs, which either provide an over- or underestimation of technical efficiency.