The mechanism of diuresis during the 1st h of immersion was investigated using anesthetized dogs. Four different experiments were carried out. First, left atrial transmural pressure was measured before, during, and after immersion. The data suggest that, although the left atrium may or may not be stretched depending on the conditions of immersion, the amount of diuresis is independent of the amount of left atrial stretch, and therefore a causal relationship between diuresis and left atrial stretch could not be established. Second, bilateral cervical vagotomy was carried out. Immersion diuresis sometimes occurred despite this vagotomy, suggesting that the left atrial stretch reflex was not participating in those cases. Third, negative-pressure breathing was carried out to simulate the negative transthoracic pressure associated with uncompensated immersion. The average left atrial transmural pressure did not change. A slight hemodilution and a moderate diuresis occurred. There was no correlation between changes in left atrial transmural pressure and changes in urine ouput. Fourth, blood studies were done on splenectomized dogs subjected to immersion. Hemodilution occurred and was most marked in dogs which had had their kidneys removed. The hemodilution is sufficient to explain the early phase of the immersion diuresis. The data suggest that, in anesthetized dogs, hemodilution is the probable initiator of diuresis upon immersion and that, in dogs, left atrial stretch is unrelated to diuresis during immersion or negative-pressure breathing.