Lake productivity is fundamental to biogeochemical budgets as well as estimating ecological state and predicting future development. Combining modelling with Earth Observation data facilitates a new perspective for studying lake primary production. In this study, primary production was modelled in the large Lake Geneva using the MEdium Resolution Imaging Spectrometer (MERIS) image archive for 2002–2012. We used a semi-empirical model that estimates primary production as a function of photosynthetically absorbed radiation and quantum yield of carbon fixation. The necessary input parameters of the model—concentration of chlorophyll a, downwelling irradiance, and the diffuse attenuation coefficient—were obtained from MERIS products. The primary production maps allow us to study decennial temporal (with daily frequency) and spatial changes in this lake that a single sample point cannot provide. Modelled estimates agreed with in situ results (R2 = 0.68) and showed a decreasing trend (∼27%) in production in Lake Geneva for the selected decade. Yet, in situ monitoring measurements missed the general increase of productivity near the incoming Rhône River. We show that the temporal and spatial resolution provided by satellite observations allows estimates of primary production at the basin-scale. The phytoplankton annual primary production was estimated as ∼302 (SD 20) g C m−2 yr−1 for Lake Geneva for 2003 to 2011. This study demonstrates that maps of primary production can be obtained even with reduced resolution (1200  m) MERIS data and relatively simple methods, and thereby calls for deeper integration of remote sensing products into conventional in situ observation approaches.