Following implementation of the tenth revision of the International Classification of Diseases (ICD) in 2000, it has become very difficult in France to reconstruct consistent cause-of-death time series based on this most recent version of the ICD. Not only are changes both numerous and complex, but in the same year France radically modified its method of producing cause-of-death statistic by replacing manual coding with a new automatic system. It is now practically impossible to distinguish between statistical discontinuities due to changes in the content of ICD items and those arising from the automatic coding procedure. Although INSERM has taken the worthy initiative of double coding a sample of deaths to ICD-9 and ICD-10, this study is, alas, of limited utility due to the very small number of observations concerned. Can we rely more successfully on the larger-scale comparability studies conducted in the United States or England and Wales, or on an in-depth analysis of multiple causes of deaths? Nothing is less certain.