Considerable controversy has followed the recent publication of studies indicating that central pontine myelinolysis is caused by rapid correction of hyponatraemia. Alternative suggestions have been that myelinolysis is due to uncorrected hyponatraemia, that it occurs only with over-correction of hyponatraemia or that it is due to coincidental hypoxia. The following experiments were undertaken to clarify the relationship between myelinolysis and derangements of serum sodium and their treatment. Severe hyponatraemia ([Na+] ⩽ 122mmol/l) was produced in three groups of rabbits by injection of vasopressin and 5% dextrose in water. Rabbits with severe uncorrected hyponatraemia sustained for seven days or more did not show myelinolysis at autopsy. Myelinolytic lesions did develop in 3 of 7 rabbits in a second group in which corrective infusion of hypertonic saline was administered after only three days of severe hyponatraemia. Neurological deterioration also occurred in rabbits in the third group which received hypertonic saline within 24 h of developing severe hyponatraemia. In this group no lesions were apparent at autopsy. No animal became hypernatraemic with correction. These results indicate that even prolonged severe hyponatraemia does not lead to myelinolysis if it remains uncorrected, but that rapid correction of hyponatraemia, without over-correction, can cause neurological disease. Accordingly, a rapid rise in serum sodium should be avoided.