Excess mortality, i.e. more deaths than expected in a similar normal population, has been reduced in Parkinson's disease by levodopa. A California series and a large collaborative study throughout the United States show the ratio of observed to expected deaths to be normal. In contrast, a New York study was similar patients and a Montreal group of much more severely disabled patients showed greater mortality. The reasons for these differences are unclear, but some variations in results may be attributed to different methods of calculating expected deaths to derive the ratio of observed to expected deaths.