In order to determine the optimal diluent for the determination of Mg by atomic absorption spectrophotometry, the following diluents were tested: deionized water, 0.1 mol/l HCl, 2.5 g/l SrCl2, 1.8 g/l LaCl3, 17.7 g/l LaCl3, half-saturated 8-hydroxyquinoline (oxine), 40 g/l trichloroacetic acid (TCA)/17.7 g/l LaCl3, and 100 g/l TCA/0.1 mol/l HCl. The calibration curves, obtained on the same day, were passed through the origin in the form of linear regressions. The variances about the calibration curves did not differ significantly (p greater than 0.05), while the calibration slopes for the diluents differed significantly (p less than 0.001). A standard serum solution was analyzed with the use of the eight diluents and the mean results did not differ significantly (p greater than 0.05). The % coefficient of variation varied from 0.8 to 2.0 and the percent recovery ranged from 95.5 to 102.5. The 99% confidence interval (CI) ranged from 0.02 to 0.09. Since conventional CIs do not take into account the random error of the calibration curve, the inverse CI were also calculated for each diluent. On both the conventional estimates and on the inverse CI, strontium chloride gave the best results, but only marginally in view of the interassay variability of approximately 1%.