Since our unexpected discovery of 12 compact radio sources in one of the nuclei of Arp220 we have been monitoring the structure using global VLBI and the VLBA. With several epochs of data we can now demonstrate that the fluxes of the radio sources are, in general, decreasing sytematically but not smoothly. The sources also have steep spectra. Both facts support our earlier supposition that these are highly luminous radio supernovae. We initially predicted a radio supernova rate of approximately 2yr−1 Our most recent observations indicate that the occurrence rate of luminous RSN is several times lower than predicted, thus our model of the Arp220 starburst must be revised.