OBJECTIVE: Elderly (65y+) are steadily growing, but this population segment is also the one in whom mortality, morbidity and health care costs increase sharply with age as a result of co-morbidity and greater frailty. This project intends to document the implications of using different modelling approaches on the benefit evaluation of a public health intervention in elderly. METHODS: We designed a mathematical model to simulate the effect of a hypothetical public health intervention aiming at reducing mortality in the 65 y+. The simulation is run on an elderly population of 1,000,000 individuals (age weighted average of 75.66 y). The impact of the intervention is compared between a cohort model (i.e., average parameters applied to the 75.66 year- old elderly cohort) and a population model (i.e., age-specific parameters applied to the entire elderly population). Life-expectancy gains (LEG) from both approaches were computed between intervention and no-intervention. Various scenarios were compared through a range of different mathematical specifications of age-specific intervention coverage and mortality reduction. RESULTS: In the cohort approach, life expectancies were respectively 11.38 and 11.48 years between no-intervention and intervention, i.e. a LEG of 0.10 y for the 75.66 y-old elderly cohort. In the population approach, age-specific life expectancies averaged 11.51 and 12.19 y between no-intervention and intervention, respectively. This translated into a weighted average LEG of 0.52 y, i.e. a gain 5-times higher than in the cohort approach. This result was confirmed in various scenarios. CONCLUSION: Population modelling, whilst being potentially more data-hungry and mathematically demanding, allows for more comprehensive consideration of age-specific parameters in the decision-making process. This approach has the potential to better capture the whole benefit of a population-wide intervention which is particularly insightful in the elderly for whom mortality, disability and costs of health care are even more age-sensitive.