A method is described for the determination of mercury by potentiometric stripping analysis. The analyte, Hg2+, is employed for oxidizing a fixed amount of cadmium, previously reduced and amalgamated at a thin mercury film preplated on a glassy carbon electrode. The cadmium stripping signal correlates well with the amount of Hg2+ added. Correlation coefficients of 0.9971 and 0.9960 were obtained for the two working ranges: (25 ng-2.5 mu g) and (5.0-50) mu g Hg2+, respectively, in spiked water samples. The method was investigated with respect to precision and accuracy by spiking a natural water sample with 25 mu g Hg2+. Nine replicate determinations gave a mean value of 24.8 mu g with a standard deviation +/-0.31 mu g. The 95% confidence limit of the mean suggested the absence of systematic errors. Using the highest possible sensitivity, detection limits of 2.0 ng (167 ng/l) and 0.5 ng (4.2 ng/100 ml of whole blood) were obtained in water and blood samples, respectively. The applicability of the method was successfully extended to include the more complex matrices after recording a zero blank from authentic samples spiked with Cd2+ (25 mu g). The described PSA procedure is a simple and rapid method compared with the cold-vapor technique, with a 5.2% and 4.9% RSD, respectively.