A two-year experiment was carried out in the Sele river Plain (Southern Italy) in order to evaluate the effects of water deficit or water excess on the performance of"Burley C103". Nine irrigation treatments were compared : 1) non irrigated control (A); 2) treatment irrigated throughout the growing cycle on the basis of ETm calculated from class A evaporation (SO); 4-5-6) treatments with no irrigation for a three week period during the post-transplanting stage (S1) or the development stage (S2) or the flowering stage (S3); 7-8-9) treatments with twice the water of S0 in the same periods as the proceding ones (D1, D2, D3 respectively). Compared to S0, giving better quantitative (3,97 t/ha) and qualitative (34% of grade A) results, water deficit in S1 induced no significative yield change, while water deficit during S2 or S3 stages decreased the yields (-16% for S2 and -15% for S3). Doubling water throughout the cycle or in each stage, affected the quality negatively. Total alkaloid content decreased with normal irrigation S0 (2,69% versus 3,40% of non irrigated control), while it showed a trend to increase with water deficit during S2 or S3; doubling water during S2 resulted in the lowest total alkaloid content (1,59%).