The interaction between centrosomes and kinetochores was studied in multinucleate cells induced by Colcemid treatment or by random cell fusion. Except for prematurely condensed chromosomes (PCC) of the G2-phase, PCCs do not develop their own spindle area. Perhaps the maturation promoting factor (MPF) fails to activate these centrosomes. In such PCCs, the kinetochore-centrosome interaction was found to be non-specific: sometimes only a few chromosomes of a group could establish connections with centrosomes, sometimes chromosomes from the same PCC group developed microtubule (MT) attachment with different centrosomes (not the pair), and sometimes kinetochores of PCC groups failed to interact with MTs. These findings explain the abnormal mitotic behaviour of PCCs as seen in the light microscope. These PCCs develop micronuclei or normal nuclei by nuclear re-formation in telophase. All the different PCC groups revealed kinetochores with kinetochore plates. It was shown that transformation of presumptive kinetochores to a trilaminar kinetochore does not depend on nuclear envelope breakdown or on the degree of chromosome condensation. This may be induced by the MPF which may initiate different events like chromosome condensation, nuclear envelope breakdown and kinetochore transformation by secondary factors. Other observations like establishment of connections by different chromosome groups to a common centrosome, kinetochore attachment of PCCs to different centrosomes, interaction of one kinetochore with two centrosomes, kinetochores being stretched and bent to receive microtubules and finally the failure of some kinetochores to develop MT attachment, all strongly suggest that the kinetochores serve as the point of termination rather than the nucleation sites of kinetochore MTs.