The reliability of electrolyte determinations with the Ektachem 700 was evaluated by various means including the use of reference method values. The influence of protein concentration, which may alter the viscosity and hence the speed of diffusion, was systematically investigated by using samples of varying protein concentration obtained by ultracentrifugation. Calcium. The mean bias between Ektachem results and reference method values of 9 control sera was -7.3%. In a comparative study with native sera a negative bias was not obtained. The accuracy was independent of the protein concentration. Chloride. The mean bias was -0.1% compared with reference method values. As might be expected, the difference between chloride values with Ektachem--using a "direct" ion-selective electrode--and determinations by coulometry increased with increasing protein concentrations. Magnesium. A negative bias (-5.0%) was obtained in accuracy control with reference method values, and to a smaller extent in a comparative study with native sera. An influence of the protein concentration on the magnesium determination was not observed. Phosphate. Accuracy control by method-dependent assigned values and a comparative study with native sera showed a positive bias. Ektachem results depend on the protein concentration. At 120 g/l protein the bias was +13.6%. Potassium. The mean bias, with respect to reference method values, was -1.2%. At high concentrations of proteins of "normal" composition, Ektachem results agree with measurements by flame atomic emission spectrometry; in paraproteinaemic sera the values are higher. Sodium. The mean inaccuracy was 3.3%, compared with reference method values. The dependence on the amount and composition of the total protein was similar to that found for potassium analysis.