Studies examining the mortality from Parkinson's disease have been reviewed to assess the impact of levodopa therapy. These include national mortality data taken from death certificate returns, cohort studies comparing the deaths observed in a group of patients with those expected from population statistics, and case-control studies comparing deaths in patients with those in a matched control group. It is concluded that the decrease in crude mortality rate in the early 1970s in several western countries was mirrored by a decrease in observed to expected mortality rates in cohort studies over the same period. The rise in crude mortality in the late 1970s and early 1980s paralleled an increase in observed to expected mortality in cohort studies. These trends are compatible with the hypothesis that levodopa delayed the death of a cohort of frail elderly parkinsonian patients who succumbed approximately 5 years later, leading to an apparent "catch-up" increment in national mortality data. Whilst crude mortality in England and Wales does not appear to be increasing as fast as before the introduction of levodopa, recent cohort and case-control studies report mortality ratios comparable to those seen in the prelevodopa era. Further case-control studies are required to determine the position with current treatment regimens.