The concentrations of elements in biological tissue may be influenced by a large number of environmental and physiological factors. Even when all known sources of variability have been either eliminated or taken into account, a very high degree of unexplained residual variability may persist between individual organisms within the same population. In the present study, a simple statistical method is described which permits the calculation of the residual variabilities of element concentrations, even from complex multivariate data, and allows statistical comparison between elements to be made (either within a single tissue, between different tissues or between different species). The method is quite general and could also be used for studying residual variability in any other natural phenomena (e.g. enzyme concentrations, water temperature). The case of 25 element concentrations in the whole soft tissue of the mussel Mytilus edulis collected from Bellevue, Newfoundland, Canada in June 1988 was used as a model. It was clearly shown that some elements (e.g. the alkali earth elements and B, Mg and Cu) had extremely low residual variability while other elements (e.g. Ce, Zn, Ba, La, U, Pb, Ag, Y, Sr and Ca) showed unusually high degrees of residual variability. Al also showed very high variability but this appeared to be due to the presence of undigested sediment in the gut rather than to residual variability. Important sources of non-residual variability included sex, size and growth rate. The method described in this paper could be used as an initial screening test for pinpointing intrapopulation genetic differences in element metabolism between individual organisms.