In using ratios of pairs of elements, such as strontium-calcium and cesiumpotassium, to predict contamination levels, it is necessary to relate the ratio in the donor material to that in the acceptor material by means of a coefficient-the discrimination factor. If a constant number for any one discrimination factor relating a donor-acceptor system is used, errors can result because the discrimination factor can change with the degree of mixing of the two elements, with the environment the elements occur in as they move from one link of the biological chain to the next, and with time. Whether these changes are of practical significance in radiation hazard control, particularly where only averages are needed, is unknown. However, experimental and theoretical evidence indicates that variations in discrimination may occur under certain conditions in the field, and thereby could affect the confidence given to predictions of future contamination levels based on current measurements: This paper describes experimental and theoretical studies related to the variability of discrimination, suggesting the need for caution in using ratios, and recommends that contamination levels be reported in terms of conventional units as well as ratios.