Famines are extreme cases of environmental stress, and have been used by a series of studies to explore the long-term consequences of the fetal or childhood environment. Results are inconsistent and do not support negative long-term effects on mortality. The authors test the hypothesis that selection during famine changes the frailty distributions of cohorts and may hide negative long-term effects. They use death counts from age 60+ from the Human Mortality Data Base for the birth cohorts 1850-1854, 1855-1859, 1860-1865, 1866-1868, 1869-1874, 1875-1879, 1880-1884 and 1885-1889 to explore the effect of being born during the Great Finnish Famine 1866-1868. Swedish cohorts without famine exposure are analysed as a control group. Cohorts born in Finland during the Great Finnish Famine are highly heterogeneous in their distribution of deaths after age 60. By contrast, cohorts born in the years immediately after the famine are particularly homogeneous. Accounting for these differences results into a lower remaining life expectancy at age 60 for cohorts born during the famine. Statistically, long-term effects of famine on mortality become only visible when changes in the frailty distribution of cohorts are explicitly considered.