INTRODUCTION AND AIMS: In the 1990s, a strong inverse relationship between life expectancy (LE) in Russia and mortality from alcohol poisoning was observed. This association is remarkable as this cause accounts for less than 2% of deaths each year. It can be explained by treating the alcohol poisoning mortality as the best available measure in Russia of the population prevalence of harmful drinking in any year which in turn associated with mortality from a wide range of causes. This study analyses the evolving relationship of LE with this mortality-based measure of harmful drinking since 1965, and places it in a policy context. DESIGN AND METHODS: We examine three periods: 1965-1984, a period of gradual LE decline; 1984-2003, a period of massive LE fluctuations; and 2003-2017, a period of LE improvement. Pearson's correlation coefficients and a linear relationship between annual changes in LE and alcohol poisoning were estimated for each period. RESULTS: The strongest negative correlation between changes in LE and alcohol poisonings was found in 1984-2003. Over the period 2003-2017 a consistent positive LE trend emerged that was statistically independent of alcohol poisoning. DISCUSSION AND CONCLUSIONS: These results suggest that in the period from the mid-2000s a growth of LE in Russia was to a large extent independent of changes in the population prevalence of harmful drinking. While there has been a reduction in mortality at ages 15-64, at older ages mortality reduction unrelated to alcohol has become an increasingly important driver of overall mortality improvements.