Whether or not an increased serum level of an unmeasured cation will reduce the anion gap (AG) depends on concomitant changes (or lack thereof) in serum unmeasured anions. In the present retrospective study, we sought to determine the effect of lithium carbonate or citrate and magnesium sulfate on the AG. Two groups of psychiatric patients whose average serum lithium levels were 0.6 and 1.0 mEq/L were studied. The AG in each group (10.2 +/- 0.3 [SE] and 9.0 +/- 0.4 mEq/L, respectively) was significantly (P less than 0.05) lower than that of a control group (11.1 +/- 0.3 mEq/L). Separately, we collected 87 pairs of AG and serum magnesium data of patients with toxemia of pregnancy. These included those of 15 patients evaluated both before and during magnesium sulfate infusion. Despite an average serum magnesium level of 4.1 +/- 0.2 mEq/L, the AG of the latter subjects tended to increase slightly from 10.8 +/- 0.5 to 11.7 +/- 0.7 mEq/L during magnesium infusion, instead of decreasing. The other data (72 determinations in the 55 patients without preinfusion values) revealed a mean AG of 11.0 +/- 0.3 mEq/L, coexisting with an elevated serum magnesium of 4.1 +/- 0.1 mEq/L. None of the above-mentioned three mean AG values differed significantly from that of the control group. We conclude that hypermagnesemia resulting from the administration of magnesium sulfate does not reduce the AG, probably because of a concomitant and proportional increase in serum sulfate. The present data reemphasize the need to assess the anion gap as the result of concomitant changes in both unmeasured anions and cations.