The epidemiological transition has provided the theoretical background for the expectation of convergence in mortality patterns. We formally test and reject the convergence hypothesis for a sample of industrialized countries in the period from 1960 to 2008. After a period of convergence in the decade of 1960 there followed a sustained process of divergence with a pronounced increase at the end of the 1980's, explained by trends within former Socialist countries (Eastern countries). While Eastern countries experienced abrupt divergence after the dissolution of the Soviet Union, differences within Western countries remained broadly constant for the whole period. Western countries transitioned from a strong correlation between life expectancy and variance in 1960 to no association between both moments in 2008 while Eastern countries experienced the opposite evolution. Taken together, our results suggest that convergence can be better understood when accounting for shared structural similarities amongst groups of countries rather than through global convergence.