Because the primary stimulus for thirst is an increase in plasma tonicity, we hypothesised that dehydrated horses would drink a greater total volume of fluid voluntarily during the first hour of recovery when they were initially offered salt water. To test this hypothesis, bodyweight (bwt), fluid intake (FI) and [Na+] were measured in 6 Arabian horses offered 3 rehydration solutions. After dehydration was induced by frusemide administration (1 mg/kg bwt, i.v.) followed by 45 km treadmill exercise, water (W), 0.45% NaCl and 0.9% NaCl were offered, in a randomised order, during the initial 5 min after completing exercise. Horses were subsequently placed in a stall and further intake of plain water during the first hour of recovery was measured. By the end of exercise, horses lost 5.2 +/- 0.2, 5.6 +/- 0.3 and 5.7 +/- 0.2% (P>0.05) bwt and FI during the first 5 min of recovery was 10.5 +/- 0.7, 11.6 +/- 0.8 and 11.6 +/- 1.5 l (P>0.05) for W, 0.45% NaCl and 0.9% NaCl, respectively. After 20 min of recovery, [Na+] had decreased with W but remained unchanged from the end exercise values for both saline solutions. During the initial hour of recovery, further water intake was 0.9 +/- 0.4, 5.0 +/- 0.5 and 6.9 +/- 0.7 l (P<0.05) for W, 0.45% NaCl and 0.9% NaCl, respectively. Therefore, total FI was 11.4 +/- 0.5, 16.6 +/- 0.7 and 18.5 +/- 1.7 l (P<0.05) for W, 0.45% NaCl and 0.9% NaCl, respectively, and persisting bwt loss after 60 min of recovery was greater (P<0.05) for W (3.5%) than for the 2 saline solutions (24% for 0.45% NaCl and 1.9% for 0.9% NaCl). In conclusion, providing salt water as the initial rehydration fluid maintained an elevated [Na+] and resulted in greater total FI and recovery of bwt loss during the first hour of recovery, in comparison to offering only plain water.