Abstract During the second half of 1997, JET carried out a broad-based series of D–T experiments (DTE1) producing a total of 675MJ of fusion energy. A large scale tritium supply and processing plant, the first of its kind, allowed the repeated use of the 20 g tritium on site, supplying a total of 99.3 g of tritium to the machine. After DTEl, the tritium inventory in the torus remained a factor of about three higher than expected from the Preliminary Tritium Experiment in 1991, and this is thought to be related to tritium-saturated carbon films on surfaces which are shadowed from erosion by the plasma. During DTE1 records were set for peak (16.l MW) and quasi steady-state (4 MW for 4 s) fusion power and for the ratio of fusion power to input power (0.62; if a similar plasma could be obtained in steady-state, the Q would be 0.94 ± 0.17). Alpha particle heating was clearly demonstrated and shown to be consistent with classical expectations. In the optimised shear mode of operation internal transport barriers were established for the first time in D–T, with a threshold power roughly equal to that in D–D. ELMy H-mode studies in D–T have considerably strengthened the basis for predicting the heating requirements and performance of ITER. Candidate ICRF heating schemes for ITER were successfully tested and the relevant simulation codes validated. With regard to isotope effects in ELMy H-modes, the ITER scaling for the H-mode threshold power had to be modified to include an inverse mass dependence (≈ A −1 ), while energy transport showed little dependence on isotope and seems to involve different physics in the edge and the core of the plasma. JET confinement data obtained under conditions which were identical to ITER in most dimensionless parameters scale close to gyro-Bohm and predict ignition for ITER provided the required densities can be reached.