The paper highlights the JET work in physics and technology during the period of the JET Joint Undertaking (1978-1999), with special emphasis on what has been learned for extrapolation to a NEXT STEP device. - Global confinement scaling has been extended to high currents and heating powers. Dimensionless scaling experiments of ELMy H mode plasmas suggest that bulk plasma transport is gyro-Bohm and predict ignition for, a device with ITER-FDR parameters. Experiments in which the plasma elongation and triangularity were varied independently show a strong increase of confinement time with elongation (tau(E) similar to kappa(a)(0.8+/-0.3)) thus supporting a basic design principle of ITERFEAT. With the Pellet Enhanced Performance (PEP) mode, JET has discovered the beneficial effect of reversed magnetic shear on confinement, opening the possibility of advanced tokamak scenarios. - With a three stage programme of progressively more closed divertors, JET has demonstrated the benefits of divertor closure, in particular, of high divertor neutral pressure which facilitates helium removal. It has also shown that. in detached (or semidetached) radiative divertor plasmas the average power load on the target plates of a NEXT STEP device should be tolerable but, in addition; that the transient. power loads during ELMS could cause problems. - In 1991 JET has demonstrated the first ever controlled production of a megawatt of fusion power. More extensive D-T experiments in 1997 (DTE1) have established new records in fusion performance: 16 MW transient fusion power with Q(in) = 0.62 (i.e. close to breakeven, Q(in) = 1) and 4 MW steady state fusion power with Q(in) = 0.18 for 4 s. DTE1 has also allowed a successful test of various reactor ICRF heating schemes and a. clear demonstration of alpha particle heating, consistent with classical expectations. - JET has developed and tested some of the most important technologies for a NEXT STEP and a reactor, in particular the safe handling of tritium and the remote handling of large equipment as demonstrated by the Remote Divertor Exchange following DTE1.