In the current context of increase in number and geographic expansion of wild ungulates combined with the decreasing hunting pressure, wildlife managers need efficient tools for better control of populations. In general, managers use theoretical maximal growth rates of population and/or abundance estimates to set hunting plans. Concerning the use of population growth rates, these can vary greatly depending on the year, the status of the population (e.g. in the colonisation phase), or others factors, and therefore lead to poor management decisions. When relying on population abundance only does not, however, give a comprehensive picture of the population functioning. To become more insightful, abundance needs to be complemented by additional biological measurements such as individual performance (e.g. body mass) or the browsing pressure. The complementary use of several aspects of the population-environment system is the heart of indicators of ecological change (IEC) proposed to manage large herbivore populations. Although the use of IEC makes it possible to improve ungulate management by adjusting quotas upwards or downwards according to the its temporal trends, the numerical adjustment of hunting plans is based on expert opinion. To achieve more objective decisions, we suggest to combine IEC, hunting data and demographic parameters from long-term monitoring of two red deer populations into a demographic model and to predict past and future population trajectories to get a reliable of estimate of the annual population growth rate. We here developed an age-structured matrix model incorporating harvest data to predict population abundance variation and to improve the setting of annual hunting quotas (quantitatively and qualitatively by age and sex classes). We used longitudinal observation of animals of two red deer populations to parametrize the model with empirical demographic rate estimates: La Petite Pierre and Chambord (France) along with the corresponding harvest data. As expected for a long-lived species, natural survival of females showed little variability among years. Conversely, pregnancy rates of females varied with age class (yearling vs adult) and between years in a density-dependant manner. From these demographic rates, we could back-cast population trajectories using relative abundance time series from spotlight counts (IEC), to force simulations within biologically realistic boundaries. Finally, we used this model to predict future population trajectories and propose hunting quotas according to the management objectives. Since 2013 at la Petite Pierre and 2015 at Chambord, we successfully experimented this approach to help managers to both decrease abundance population and improve demographic performance.