The accuracy and reproducibility of two different Syva EMIT and Abbot TDx instruments for the determination of gentamicin were assessed. A series of spiked standard serum samples containing gentamicin in concentrations ranging from 0.88-17.82 mg/l were used for this purpose. Both TDx systems evaluated had excellent reproducibility but overestimated gentamicin levels in the clinically important concentration range in an apparent concentration-dependent way by at least 5 and 10%, respectively. In addition both these systems exhibited considerable carryover effects when high concentration gentamicin samples (17.82 mg/l) were alternated with serum blanks. Average gentamicin levels determined by the EMIT systems were much closer to the target values, especially those of EMIT-A, than those measured by the TDx-systems. The reproducibility, as measured by the coefficient of variation (CV), of the semi-automated EMIT systems over the entire concentration range investigated was about half that of the TDx-systems. No carryover effects were observed with the two EMIT-systems. This study has shown that the accuracy and reproducibility of both assay methods vary considerably over the clinically important gentamicin concentration range. In addition there may be a large variation between instruments of the same system especially in the critical trough region for gentamicin.