Résumé : This paper presents the experience of new generation of telemedicine systems (called DIATELIC) which includes a self-learning "intelligent" expert system detecting hydration disorders. From 2002 each patient starting PD and living in Lorraine (East of France, total population 2.300 000 w) is equipped with a computer station through which he connect himself to a database, to record his daily parameters: weight, pro and decubitus blood pressure, ultrafiltration and tonicity of the dialysate. All the connections are established by secure-internet thanks to a standard web navigator. An integrated e-mail system allows to improve patient-doctor communication. A total of 94 patients studied during a 3-years period. During the study, a total of 3.650 messages were sent and processed. 90 % of them where patient-doctor messages, and 10 % were doctor-doctor messages. The mean delay for response to a patient was 26 hours and to a doctor 22 hours. The hydration status of the PD patients was improved: during the study the mean weight decreased from 66 to 65 Kg and the mean blood pressure decreased from 10 to 9 mmHg. Previously to the study, the mean hospitalization rate of the PD patients was 19 days/year/patient. It showed a dramatic improvement to 10 days/year/patient during the study. The causes of hospitalizations were 60 % not PD related, and only 5 % due to hydration disorders. The large use of telemedicine with PD (DIATELIC) resulted in a dramatic decrease in the global hospitalization rate (and then the global cost of the therapy) and improves the PD patient's follow-up. For users (patients and doctors) it appears now to be as indispensable as the phone.