This prototype climate service uses a large dataset of daily meteorological and mortality data from 147 European regions in 16 countries, which represent a total population of more than 400 million people. Associations have been calculated by means of Distributed Lag Nonlinear Models (DLNM), controlling for seasonality and long-term trends, and including a range of lags for the modelling of delayed effects. The system is driven by weather forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF) Atmospheric Model Ensemble 15-day forecast (ENS) for lead times of up to 15 days, addressing the current gap in existing systems in terms of the predictive capacity for lead times beyond a few days. Health forecasts have been generated combining these forecasts of temperature, the temperature-mortality associations in each region and the daily mortalities. The final result comes from the study of the predictability of the temperature-attributable mortality and its decay as a function of the lead time.