The effects of an endurance flight on the haematocrit, the percentage of packed red blood cells per blood volume, were examined within the framework of six possible factors explaining possible changes in the haematocrit. Two approaches were adopted: (1) the haematocrit was studied in four species of passerine birds which landed on an Italian island after having crossed the Mediterranean Sea on their spring migration in a non-stop flight; (2) the haematocrit was evaluated in six individual red knots after a flight of 1, 2, 4 and 10 h in a wind tunnel and the data thus obtained compared with data on resting birds with or without food. In the four passerine species, the haematocrit decreased from 51% in fat birds to 48% in lean birds. In lean birds, the haematocrit dropped from 48% in birds with well-developed breast muscles to 36% in birds with emaciated breast muscles. In the red knots, the haematocrit was dependent on body mass in flying and resting birds. The haematocrit decreased from about 51% pre-flight to about 49% within 1 h of flight and remained at this level for up to 10 h of flight. Taking the results from the passerines and the red knots together, it seems that the haematocrit drops by a few percentage points within 1 h after the onset of flight, decreases very slowly with decreasing body mass and decreases more steeply in very lean birds having entered stage III of fasting. This indicates that dehydration is not an underlying factor in decreased haematocrit because if this were the case we would expect an increase with endurance flight. We found no effect of the presence of blood parasites on haematocrit. With the onset of flight, haemodilution may be adaptive, because it reduces blood viscosity and, thereby, energy expenditure by the heart, or it may be a sign of water conservation as an insurance against the risk of dehydration during long non-stop flights. During endurance flight, a reduction in the haematocrit may be adaptive, in that oxygen delivery capacity is adjusted to the decreased oxygen needs as body mass decreases. A decreasing haematocrit would also allow birds to reduce heart beat frequency and/or heart size, because blood viscosity decreases disproportionally with decreasing haematocrit. However, when energy stores are about to come to an end and birds increase protein breakdown, the haematocrit decreases even further, and birds probably become anaemic due to a reduced erythropoiesis.