The reduction of copper oxide species dispersed in microporous and mesoporous matrices has been studied by TPR, XPS/XAES, and XAFS. While the reduction of bulk CuO and of Cu(II) in mesoporous MCM-48 (5.6 wt-%) proceeded in one step without intermediate Cu(I) being detectable under the experimental conditions, Cu(II) in microporous matrices was reduced ill two steps. The two-step scheme cannot be identified with the reduction steps Cu(II) -> Cu(I) and CL(I) -> CU(0). Instead, highly disperse Cu(0) may be present already after the first reduction step. In siliceous matrices, coexistence of Cu(0), and Cu ions was observed over a wide temperature range, obviously due to the absence of ail autocatalytic reduction process. The latter occurred in Cu-ZSM-5, apparently involving simultaneous segregation of Cu metal from the matrix. This suggests that very small (oligorneric) Cu metal Clusters are unable to activate hydrogen. The reduction behaviour of Cu in Y zeolite depends critically on the thermal history of the sample due to the population of hidden sites by copper upon calcination. Highly disperse Cu particles are stable in MCM-48 up to 500 degrees C.