Anthropologists are constantly seeking to improve methods for age estimation in the human skeleton. A new method was introduced about a decade ago that assesses the morphological changes that take place in the acetabulum as an individual ages. The pelvis is usually well preserved in forensic cases, which makes this method potentially valuable as an adult age indicator. This method employs seven variables, each with its own set of phases. To test the accuracy and reliability of this method, 100 black South African male acetabula from the Pretoria Bone Collection were assessed based on the criteria described in the original study. Box plots and transition curves were constructed to establish whether progression with age was visible and how it could possibly be modelled. Inter-observer reliability was also assessed by making use of Fleiss’s Kappa statistic. Five specimens were used as out-of-sample examples for which maximum likelihood (point) estimates were calculated. The results demonstrated that middle and older individuals’ age estimates were vastly underestimated. Inter-observer repeatability was poor, which suggested that the classification system most likely needs to be modified. A discussion and recommendation is given for improvement of reliability and repeatability of this method.