The base-metal and PGE contents of samples from magmatic sulfide mineralization are commonly correlated with their sulfide contents, indicating that the metal contents of bulk sulfides remain approximately constant within a given prospect or part thereof. Calculated sulfide metal contents provide valuable information in mineral exploration and research, but there are few formal descriptions and analyses of the procedures. Sulfide metal contents are best calculated using an assumed value (35.7% S) for a typical pyrrhotite-chalcopyrite-pentlandite mixture, and there appears to be little advantage in accounting for sulfide species separately. Regression of metal data against sulfur is probably the most rigorous approach, but is not always practical. Above 10% S, calculations are very robust, but lower sulfide contents generally demand at least some correction for non-sulfide-hosted metals. Such corrections can become significant below 5% S, and/or in olivine-rich samples. They are best accomplished by mass-balance calculations, using concentration data from unmineralized host rocks. Significant uncertainties are introduced by analytical errors for sulfur, base-metals, and PGE, which are commonly measured from separate sample aliquots. These combined errors in sulfide metal contents generally exceed ±10%, but expand further at low S contents. In general, treatment of data from samples containing <2.5% S must be approached with caution, especially for PGE, for which the exact host minerals may not be known. Application of the method in simple grade-potential assessment is straightforward, but research studies involving sulfide-poor samples are inherently more complex. Under-correction or over-correction of data for non-sulfide-hosted metals can lead to false negative or positive correlations between sulfide metal contents and sulfide content. As the latter may itself be linked to geological parameters, such as depth within an intrusive body, undue significance could be ascribed to such trends. There are also valid geological reasons for such correlations, and such data require careful assessment to separate true and artificial variations. Propagated analytical uncertainties increase significantly in sulfide-poor samples, and must also be borne in mind whenever data from different localities or units are compared and contrasted.