BACKGROUND: Measuring the impact of competing risks of death on society is important for setting public health policy and allocating resources. However, various indicators may result in inconsistent conclusions. The potential gains in life expectancy (PGLE) by elimination of deaths from HIV/AIDS, diseases of the heart and malignant neoplasms were compared to the years of potential life lost (YPLL) due to these causes in measuring the impact of premature death for the US population of working age (15-64 years). METHODS: The PGLE and the YPLL were computed from mortality reports (1987-1992) by race and gender group for deaths from HIV/AIDS, diseases of the heart and malignant neoplasms for the US population of working age. RESULTS: The YPLL overestimated the importance of premature deaths from HIV/AIDS compared to the PGLE. For the total US population and total US white population of working age, the YPLL were about 20-30% higher than the PGLE. However, the YPLL were about 20-30% lower than the PGLE for the US black population of working age. Furthermore the relative importance of the impact of death from various diseases may be interchanged by these two indicators. For example, for US black males of working age, the impact of deaths from HIV/AIDS by PGLE in 1992 was higher than that from malignant neoplasms and lower than that from diseases of the heart, but by using YPLL, the impact of premature deaths from HIV/AIDS was higher than that from both diseases of the heart and malignant neoplasms. CONCLUSIONS: The PGLE by elimination of deaths from diseases takes into account the competing risks on the population and it can be compared easily across populations. The YPLL is an index that does not take into account competing risks and it is also heavily influenced by the age structure and total population size. Although there are several standardization techniques proposed to improve the comparability of the YPLL across different populations, the YPLL fails to address the central issue of competing risks operating on the population. For this reason, we prefer the PGLE to the YPLL in measuring the impact of premature deaths on a population.