In 30 obese men, calcium, magnesium, and phosphate balances were measured for 40 days using one of four weight reducing regimens: 1) 400 kcal soy protein; 2) 400 kcal collagen protein; 3) total fasting with potassium; and 4) total fasting without potassium. Relationship of the minerals to each other and to nitrogen and to the QTc interval was also examined. All groups were in negative cumulative calcium balance but the protein-fed groups lost less calcium (soy, -3.0 +/- 2.1 g; collagen -4.9 +/- 3.2 g) than the total fasting groups (with potassium supplement, -9.2 +/- 3.4 g; without potassium supplement, -5.8 +/- 2.1 g) (p less than 0.01). The soy-fed group attained positive cumulative magnesium (0.7 +/- 0.5 g) and phosphate balances (6.9 +/- 3.9 g). The other three groups had significantly more negative magnesium (p less than 0.0005) and phosphate (p less than 0.0005) balances, (collagen, magnesium balance, -1.1 +/- 1.0 g, and phosphate balance, -7.6 +/- 3.7 g; total fasting without potassium, magnesium balance, -1.4 +/- 0.6 g, and phosphate balance, -5.4 +/- 2.7 g). Potassium supplementation during fasting increased urinary losses of calcium and fecal losses of magnesium. High phosphate intake reduced urinary calcium. Nitrogen losses predicted only magnesium losses. Serum mineral levels did not reflect tissue mineral status. Shortening in the QTc interval as an indicator of reduced myocardial instability was related to the increase in serum phosphate in the protein-fed subjects.