Compared with the findings for an age-matched group not taking deprenyl, a higher risk of mortality in Parkinson's disease patients taking deprenyl has recently been reported. Since a biological basis for this observation was not apparent, an epidemiological explanation was sought. Expected mortality over a 6-year period in four hypothetical age-matched groups was determined. Although groups were age matched, ages of individuals within the groups varied. Variation of individual ages within each group, without affecting the age-match comparability, produced a marked variation in expected group mortality. Mortality comparisons between age-matched groups can be invalid. This epidemiological trap might account for the recent unexplained high mortality observed in a group of Parkinson's disease patients taking deprenyl.