This study completes a previous one in which the authors tested the electrogustometric threshold in 300 normal subjects used as controls. The collected datas has been computerized on a programable calculator according to a lognormal model. Comparison between the electrogustometric threshold in diabetic patients, ethylic ones and controls demonstrate that they are statistically different. More than a half of the diabetics and ethylics have abnormaly high electrogustometric thresholds. The responses have been studied as a function of the evolution of the disease, of the clinical form and of gravity. Diabete and chronic ethylism do not disturb the gustatic function at the same speed and with the same intensity: one year evolution is at least necessary for diabetics, and ten years for ethylics in order to detect abnormal electrogustometric thresholds. But if the action of ethylic intoxication is slower, it determines the most important gustometric deteriorations.