1. Three oral glucose tolerance tests were performed in each of 32 symptomatic postprandial hypoglycaemic patients (before placebo, before doxepin therapy and after doxepin therapy). Plasma neurotransmitters were determined in parallel with assays of plasma insulin and glucose levels. 2. Three different types of patients were distinguished. Type I showed a low noradrenaline/adrenaline ratio, high dopamine levels and low platelet 5-hydroxytryptamine (serotonin) levels during basal periods. After a glucose load, late peaks of dopamine and free 5-hydroxytryptamine, which coincided with the symptoms but not with the nadir of plasma glucose, were observed. Type II showed a low basal plasma noradrenaline/adrenaline ratio. After a glucose load, progressive increases in adrenaline and decreases in glucose were seen. Adrenergic symptoms coincided with the nadir of glucose. Although type III patients showed hyperinsulinaemia after a glucose load similar to the other types of patient, they did not show hyperglycaemia, but rather exhibited a sustained and progressive reduction in plasma glucose. These patients were characterized by a high basal plasma noradrenaline/adrenaline ratio, high basal plasma levels of 4-hydroxy-3-methoxyphenylethyleneglycol and high basal levels of platelet 5-hydroxytryptamine, all of which increased after a glucose load. Systolic and diastolic blood pressure decreases paralleled reductions in heart rate and glucose. The nadir of plasma glucose occurred simultaneously with the appearance of symptoms (weakness, heartburn, oppressive chest pain, tension headache, abdominal cramps, dizziness, etc.). Therapy with doxepin led to disappearance of the symptoms within 3-4 weeks. Normalization of all other disordered variables (cardiovascular, metabolic and neurochemical, and the clonidine test) paralleled the disappearance of the symptoms. 3. Symptoms varied in the three types of patients and we conclude that they are related to hypoglycaemia-induced disorders of plasma neurotransmitters, rather than to hypoglycaemia per se. We postulate that an uncoping stress situation (type I and II patients) and depression (type III patients) underlie the physiopathological mechanisms.