Several studies have demonstrated that tolvaptan (TLV) can improve hyponatremia in advanced heart failure (HF) patients with rare chance of hypernatremia. However, changes in serum sodium concentrations (S-Na) in patients with or without hyponatremia during TLV treatment have not been analyzed.Ninety-seven in-hospital patients with decompensated HF who had received TLV at 3.75-15 mg/day for 1 week were enrolled. Among 68 "responders", who had achieved any increases in urine volume (UV) during the fi rst day, urinary sodium excretion during 24 hours (U-NaEx(24)) increased significantly during one week of TLV treatment along with higher baseline S-Na (P < 0.05 and r = 0.325). Considering a cut-off value (S-Na, 132 mEq/L; AUC, 0.711) for any increases in U-NaEx(24), we defined "hyponatremia" as S-Na < 132 mEq/L. In hyponatremic responders (n = 25), S-Na increased significantly, although 1 week was not sufficient for normalization (125.8 ± 5.0 versus 128.9 ± 4.3 mEq/L, P < 0.05), along with unchanged U-NaEx(24) (2767 ± 2703 versus 2972 ± 2950 mg/day, NS). In contrast, in normonatremic responders (n = 43), S-Na remained unchanged (136.6 ± 3.1 versus 137.4 ± 2.9 mEq/L, NS) along with increased U-NaEx(24) (2201 ± 1644 versus 4198 ± 3550 mg/day, P < 0.05). TLV increased S-Na only in hyponatremic responders by way of pure aquaresis, but increased U-NaEx(24) only in normonatremic responders, which explains the scarcity of hypernatremia. Epithelial Na-channels in the distal nephrons, whose repression by TLV increases urinary sodium excretion, may be attenuated by reduced ATP-supply in worse hemodynamics under hyponatremia.