This study was conducted to compare the value of an older formulation of potassium-magnesium citrate (K4MgCit2) with newer formulations (K3MgHCit2 and K5MgCit2Cl) with respect to the correction of thiazide-induced hypokalemia and magnesium loss, alkalinizing effect, and citraturic action. Sixty-two healthy volunteers first took hydrochlorothiazide 50 mg/day. After 3 weeks of thiazide treatment (or earlier if hypokalemia developed), they were randomized to take one of three drugs for 3 weeks while continuing thiazide: K4MgCit2 (49 mEq K, 25 mEq Mg, and 74 mEq citrate/day), K3MgHCit2 (49 mEq K, 33 mEq Mg, and 98 mEq citrate/day), and K5MgCit2Cl (49 mEq K, 20 mEq Cl and 59 mEq citrate/day). Outcome measures were changes in serum potassium and magnesium, and urinary potassium, magnesium, pH, and citrate. The three drugs were equally effective in correcting thiazide-induced hypokalemia. K3MgHCit2 and K4MgCit2 produced a small but significant increase in serum magnesium concentration, whereas K5MgCit2Cl did not. Although all three supplements significantly increased urinary pH and citrate, these effects were more marked with K3MgHCit2 and K4MgCit2 than with K5MgCit2Cl. All three supplements were generally well tolerated, with the lowest side effect profile obtained with K4MgCit2. The new formulation of K3MgHCit2 exerts similar correction of thiazide-induced hypokalemia and magnesium loss, and enhancement of urinary pH and citrate, compared with the older K4MgCit2. However, it is less well tolerated. The new formulation of K5MgCit2Cl does not avert magnesium loss, and has less prominent alkalinizing and citraturic effects than the older preparation.