Summary. --Lifetime and quantum yield measurements concerning the fluorescence of Cu + impurity centre in KI were performed in the range 19-300 K. The emission mechanism was shown to involve two excited states in thermal equilibrium, one of them being metastable and lying close under the emitting level, ~e. according to the Pedrini model for Cu § fluorescence in alkali halides. Monovalent copper impurity in alkali halides has long been an active area of research. An interesting feature is that in several host matrices Cu + (as well as other ions) do not occupy the position of the ion it substitutes: in other words, it goes ,,off-centre,. Forecast for off-centre displacement was obtained by one of the authors[l-3], but essentially it concerned the ground-state configuration of monovalent impurity ions (Li t, F-, CI-, Na t, Ag t, Cu t) in alkali halide crystals as well as of Mn 2 t impurity ion in some oxides. In principle, the position in the lattice of Cu § in the relaxed excited state (RES) could be different from that in the ground state. For this reason, in the past Piccirilli and Spinolo[4] carried out lifetime measurements as a function of the temperature T concerning the fluorescence of the substitutional Cu § ion in several alkali halides (/.e. those not containing Li and F): it was assumed that the involved 3d94si--, 3d 1~ forbidden transition is made possible by odd lattice vibrations that dynamically mix the initial and final wave functions with different-parity wave functions. These authors, on the basis of the temperature behaviour of the lifetime as well as of the absorption oscillator strength, tried to determine[4] the change on off-and on-centre configuration of the Cu + in its RES. Nevertheless, these authors subsequently apologized [5] for an instrumental error in their lifetime measurements at low T. Consequently, Bertolaccini et al. repeated the experiment [6] by using a better technique, but restricting the investigation to NaC1, NaBr, KBr and RbBr. They noticed [6] that in some cases the lifetime dependence as a function of T is similar to that observed for the oscillator strength of Cu + in an on-centre configuration (i.e. involving a coth function) and, what is more, suggested the possibility of a change from an on-centre (at low T) to an off-centre configuration (at high T).In 1978, Pedrini suggested [7] that the temperature dependence of the Cu + 1199