The objectives of the present study were to investigate the effects of Aquatraining of horses (aqua-treadmill exercise; treadmill manufactured by Equitech - L.u.S. Equipment, Warendorf, Germany) on selected blood parameters [lactic acid concentration (mmol/l), haemoglobin content (g/l)] and on heart-rate variability (HRV) [heart rate (beats per min; b.p.m.), standard deviation of all NN-intervals (SDNN; ms), normalized power of the low and high frequency band (LFnorm, Hfnorm; au), % recurrence, % determinism and ratio(corr)]. Seven horses performed six exercise tests with different work loads (walking (x = 1.56 +/- 0.08 m/s) and trotting (x = 2.9 +/- 0.13 m/s): dry, water above the carpus and water above the elbow). The standardized test-protocol was: 5 min warm-up at walk while the water was pumped in, followed by the 20-min exercise period at walk or trot, followed by a 5-min walk while pumping out the water. Blood samples were taken prior to each test at rest in the stable, as well as exactly 5 min after the end of the 20-min exercise period. Electrocardiograms were recorded during rest and the 20-min exercise period. Compared to rest, neither the chosen velocities, the two water levels, nor the dry tests led to a significant increase of the lactic acid concentration in any horse. The haemoglobin content showed a significant increase as a result of exercise. Significant differences could be found between the heart rates at rest and the six exercise tests and between the mean of the levels 'walking' and the mean of the levels 'trotting'. An exercise-induced change of HRV was characterized by a decreasing SDNN, a significantly higher LFnorm (sympathetic influence) combined with a significantly lower HF(norm) power (parasympathetic activity) and a rising degree of order (significantly higher % determinism and nearly unchanged % recurrence) and stability (significantly rising ratio(corr)) of the recurrence plot. In conclusion, the used training-protocol for aqua-treadmill exercises only represents a medium-sized aerobic work load for horses, but the different levels of burden were indicated especially by changes in HRV.