Determining the appropriate approach to calibrating radiocarbon dates is challenging when unknown and variable fractions of the carbon sample are derived from terrestrial and marine systems. Uncalibrated dates from a large number of human skeletons from Western Cape and Southern Cape locales, South Africa (n = 187), can be used to explore alternate approaches to the marine carbon correction. The approach that estimates theoretically expected minimum and maximum values for marine carbon (“expected”) is compared to the approach that estimates observed minimum and maximum values (“observed”). Two case studies are explored, wherein skeletons interred together have non-overlapping conventional 14C ages. The case from the Western Cape is explored through carbon isotope values; the case from the Southern Cape uses nitrogen isotope values. In both cases, the approach using observed endpoints yields better date calibration results. Analysis of the large sample shows that mean values for estimated dietary % Marine, as calculated using expected and observed protocols, are significantly different. We conclude that the observed protocol is preferred, and that improved measures of the local marine reservoir (ΔR) are needed for this region.