A new version is proposed for the universal empirical formula, EPAX, which describes fragmentation cross sections in high-energy heavy-ion reactions. The new version, EPAX 3, can be shown to yield cross sections that are in better agreement with expe
rimental data for the most neutron-rich fragments than the previous version. At the same time, the very good agreement of EPAX 2 with data on the neutron-deficient side has been largely maintained. Comparison with measured cross sections show that the bulk of the data is reproduced within a factor of about 2, for cross sections down to the pico-barn range.
