The risk emanating from volcanic hazards today is the highest in recorded human history, which stresses the need for improved forecasts of volcanic activity. This thesis provides further insights in this respect by an integrated approach at the interface of petrology, geochronology and geophysical modelling, using Nevado de Toluca volcano in Mexico as a case study. Analyzing the long-term history of the volcano reveals that eruptions are preceded by a recurrent pattern of magma hybridization processes. The present state of the volcano is investigated by a new zircon age inversion technique, which shows that long-dormant volcanoes can reawake and produce large eruptions over short timescales. The question why some volcanoes erupt the same magma chemistry and others sample the entire igneous spectrum is examined by thermochemical modelling. These calculations reveal a relationship between magma fluxes, compositional diversity and the size of magmatic systems that resembles the structure of natural data.