Diluting a sample before analysis, as is required by both flame photometry and indirect reading ISE analysers, gives less relevant analytical results than measuring an undiluted sample directly. This is because dilution masks the impact that protein and lipids exert on the true activity of an ion in a clinical sample. Measuring a sample directly without dilution enables only those ions which are not bound to other materials, e.g. other ions or proteins, to be detected. The levels of these free ions are more likely to reflect true physiological status than analysis of both free and bound ions. Measuring a sample of whole blood directly without dilution can only be performed accurately if the ISE analyser reference junction is unaffected by changes in haematocrit. Calibrant compositions should be of an ionic strength similar to that of physiological samples. Residual liquid junction potentials between calibrants and samples should be minimised by careful design of the reference electrode.