Lifespan variation has been attracting increasingly greater attention as a measure of population health and mortality. Several studies have analysed periods of steady mortality decline, highlighting a strong inverse relationship between lifespan variation and life expectancy. Recent research has found that this association weakens, and even reverses, when mortality does not improve equally over age. However, to date no study has comprehensively explored the behaviour of lifespan variation when mortality increases significantly. Analysing three epidemics and two famines in Europe from the eighteenth to the twentieth centuries, we find that, during these events, relative lifespan variation increases, while absolute variation declines, and that subsequently both quickly revert to pre-crisis levels. Using decomposition techniques, we show that mortality at older ages leads to a temporary increase in absolute – but not relative – variation. Moreover, female lifespan variation is less affected by the crises than that of males, because of the higher impact of infant and child mortality on male lifespan variation. By underlining different trends of lifespan variation by sex and indicator, we offer new insight into the consequences of mortality crises. Contrary to what is often asserted, we also show that the choice of lifespan variation indicator is not always inconsequential.