The late diagnosis of pancreatic cancer, at a locally advanced and metastatic stage explains in part its poor prognosis. In this situation, three phase II trials with gemcitabine, a synthetic analogue of deoxycytidine, yielded some interesting results in terms of tolerance and efficiency. A new concept has also been depicted: the clinical benefit (defined by the level of pain, the consumption of analgesics and the general condition of the patient). These trials led to the realization of a randomized phase III study including 126 patients. The study compared the use of gemcitabine, in monotherapy, with fluoro-uracil (5FU), the standard form of treatment. There was evidence of clinical benefit in 23.8% of the patients treated with gemcitabine compared to only 4.8% of those receiving 5FU. The median and the overall survival at one year were significantly higher for the patients treated with gemcitabine (5.65 months and 18% respectively, versus 4.41 months and 2% for 5FU). The tolerance profile of gemcitabine was favourable. These have been confirmed by a large compassionate study undertaken in the United States, led by Storniolo, with more than 3,000 patients. Used in monotherapy, gemcitabine thus obtained authorization in 1996 to be released in the United States and in 1998 in Europe, to be used in the case of locally advanced or metastatic adenocarcinoma of the pancreas as first line chemotherapy. The administration regimen is 1,000 mg/m2 once a week (30 minutes i.v. infusion), for 7 weeks out of 8, then 3 weeks out of 4 in consecutive cycles. It is currently considered as the standard chemotherapy in this situation. Trials have been carried out to optimize the scheme for the administration of gemcitabine by lengthening its perfusion time. Current studies are now also being directed towards an association between gemcitabine and other cytotoxics, like cisplatinum or oxaliplatin.