Anticipating the long-term growth of healthcare expenditure is one of the components of budgetary surveillance that are regularly carried out at different levels, particularly in the context of the European Union. This projection can be made using macroeconomic models reasoning at a highly aggregated level. Yet this exercise can also be carried out by microsimulation, which offers a greater potential in terms of variants and types of results. The downside of this method is of course a certain level of complexity, as it is necessary to model the individual health status patterns and the distribution of the associated expenditures rather than average values. This article presents the first steps in creating this type of model, applied to local care expenditure. This model contains two modules. The first is an “epidemiological” module, which projects a dichotomous indicator of good/poor health obtained by cross-referencing subjective and objective health data. This indicator is assessed on the panel of the Health Care and Insurance survey (l'enquête santé et protection sociale: ESPS) running from 2002 to 2008. This panel enables the estimation of the probabilities of going from good to poor health as well as the probabilities of death, which are differentiated based on health status. These probabilities are then used to progressively age the 2008 sample, up to 2032. Once the individual health statuses have been projected, the second module simulates the expenses related to these statuses, using a sequential approach that first simulates the fact of having a non-zero expenditure, and then the level of this expenditure, if it is positive. The articulation of these two modules is illustrated by a few exploratory projections. They were designed to be applied to other basic data. They could therefore be coupled with microsimulation tools applied to other aspects of demographic ageing, primarily pensions.