<p>A log-normal size frequency distribution (SFD) of a primary diamond deposit provides a useful basis for extrapolating and interpolating an anticipated size recovery derived from bulk samples, especially where a single dominant population is present. To assess a deposit for economic potential the run-of-mine (ROM) diamond value or price is a key consideration. Traditionally this value has been achieved by obtaining for valuation a sample that provides sufficient quantity of diamonds in larger sizes. An alternative approach combines a modelled SFD with a modelled price-size relationship, particularly of the carater sizes. It was found in this study that generally the average $/ct value of a diamond size fraction is linearly dependent on size for sizes above 3 grains (gr). It was also found that the value profile within a size class larger than 0.7 ct were very similar, allowing a transformation of values of single stones to that of an equivalent 3 gr (‘grain’ is equivalent to ¼ ct stone). Bayesian modelling showed that for a 5000 stone sample, a modelling approach was more accurate than the conventional method and much less sensitive to the inclusion of a single high value stone.</p>