Heavy ion collisions at the LHC are expected to provide a qualitatively different environment from existing accelerators by creating a very hot, and therefore more clearly detectable QGP via hard initial parton scatterings that can be calculated rather precisely. Extrapolating from present results, all parameters relevant to the formation of the QGP will be more favourable: the energy density, the size and lifetime of the system, and the relaxation times should all improve by a large factor, typically by an order of magnitude compared to Pb+Pb collisions at the SPS. We expect particle densities of several thousand per unit of rapidity, a freeze-out volume approaching 100,000 fm3, and an initial energy density more than one hundred times larger than the one of normal nuclear matter. The initial temperature is calculated to be close to 1000 MeV, as compared to a value of 400-500 MeV at RHIC and about 200 MeV at the SPS. The energy densities and temperatures at LHC should be far above the deconfinement threshold, allowing us to probe the QGP in its asymptotically free "ideal gas" form. The analysis of extended strongly interacting matter at the LHC will move from the fixed target regime dominated by soft phenomena into a domain where the abundant formation of "mini-jets" with transverse momenta of a few GeV plays an essential role. The hard interaction between the primary partons will lead to a rapid production of further partons and thus a dense partonic medium. Perturbative QCD calculations can be used to construct and evaluate such parton interaction cascades; they indeed show the expected rapid rise towards thermalisation, on time scales considerably below 1 fm/c, to the extremely high initial temperatures mentioned. In order to verify that a QGP was produced and to study its properties, we need probes sensitive to the earliest and hottest stages of the medium. Such probes must be hard enough to resolve the short intrinsic length and time scales and they must be able to distinguish between dense confined and dense deconfined systems. Three such probes are currently known: Bound heavy quark resonances (quarkonia), hard jets, and thermal dileptons/photons. Only charmonia as deconfinement probes have been observed at the SPS as discussed above; for the others, higher incident energies appear necessary. The deconfinement analysis of hot and dense matter must be extended to bottonium states, which are produced at sufficient rates at the LHC. The , with its very small radius, can be dissociated only at the highest energy density attainable at LHC (of order 30 GeV/fm3), while the excited states ' and '' are comparable in radius with the charmonium resonances and will serve as important consistency checks. Hard jets probe the produced medium through the energy loss of partons passing through dense matter. The theoretical aspects of this problem were recently studied in considerable detail, leading to substantial progress in the understanding of the relevant medium properties for energy loss in a QGP. In particular, the rate of energy loss was found to depend quite sensitively on the size of the medium. Furthermore, there now are indications that the energy loss in cold nuclear matter is much smaller than that in a hot QGP. Corresponding studies for the energy loss in a hot hadronic medium are still needed. The production and subsequent attenuation of fast partons will add a crucial new penetrating probe to diagnose the nature of the strongly interacting matter produced in heavy ion collisions. The temperature of the primordial medium could be best determined through measurement of the spectra of real or virtual photons. Superimposed are the soft photons from the late hadronic stage as well as the primary Drell-Yan or hard QCD photons. Whether there is a window in transverse momentum (around one to a few GeV) to actually measure such thermal dileptons or photons depends crucially on the density of the produced system; fortunately, conditions could be quite favorable at the high energy densities predicted at LHC. Another unique feature of heavy-ion collisions at the LHC is the possibility to measure a large number of observables with very high accuracy on an event-by-event basis: impact parameter, multiplicity, particle ratios and spectra and, of particular importance, size and lifetime from interferometry. Single event analysis, currently pioneered by NA49 at the SPS, will become a precision tool at very high multiplicity. One of the important design considerations for the ALICE detector is to make full use of this opportunity. It will allow the study of correlations and non-statistical fluctuations which would otherwise be washed out when averaging over many events. Such fluctuations are, in general, associated with critical phenomena in the vicinity of a phase transition.