Single-energy and dual-energy quantitative computed tomography (CT) techniques were used to analyze test solutions that contained agar, K2HPO4, and isopropanol, which stimulated collagen, mineral, and fat, respectively. The impact of the use of peripheral and anthropomorphic (central) calibration phantoms was also studied. A 10% change in fat content was found to cause errors in the estimated bone mineral content of -10 mg/mL, -14 mg/mL, and -1 mg/mL for single-energy CT at 80 kVp, single-energy CT at 140 kVp, and dual-energy CT, respectively. In the K2HPO4 solutions, the addition of 50 mg/mL agar increased the estimated bone mineral content by 20-33 mg/mL for single-energy CT, and by 12-17 mg/mL for dual-energy CT. The best estimates were obtained with central calibration and dual-energy CT, but the estimates were still 14%-24% greater than the true values. Calibration samples that more accurately simulate actual spongiosa may reduce this source of error. Caution should be exercised in the use of quantitative CT under conditions in which either the bone marrow or collagen content is altered by disease or therapy.