Specifications for durability that are based on performance tests on concrete during the early stages of construction are finding acceptance throughout the world. In South Africa the durability index (DI) approach has been accepted into local specifications. This raises a challenge in the design of a concrete mixture for a construction project that takes account of the variability in materials and test conditions to ensure that the concrete will achieve the specification requirements when subjected to DI testing on in-situ samples. This paper proposes a statistical approach to accounting for this variability in concrete mixture design with oxygen permeability index (OPI) testing, in a manner that acknowledges the variability of both laboratory-based and in-situ sample testing for compliance. The OPI test results obtained on nominally similar concretes on actual construction projects undertaken by the South African National Roads Agency Limited (SANRAL), which included the DI-based specifications, were used to illustrate the operation of the proposed approach. The results show that the target value of OPI to be achieved in the design of concrete mixtures is influenced by the level of quality control on construction sites but that this can be managed through an appropriate statistical approach.