A predictive optimal control system for micro-cogeneration in domestic applications has been developed. This system aims at integrating stochastic inhabitant behavior and meteorological conditions as well as modelling imprecisions, while defining operation strategies that maximize the efficiency of the system taking into account the performances, the storage capacities and the electricity market opportunities. Numerical data of an average single family house has been taken as case study. The predictive optimal controller uses mixed-integer and linear programming where energy conversion and energy services models are defined as a set of linear constraints. Integer variables model the start-up and shut-down operations as well as the load dependent efficiency of the cogeneration unit. The proposed control system has been validated using more complex building and technology models to asses model inaccuracies. Typical demand profiles for stochastic factors have been used. The system is evaluated in the perspective of its usage in Virtual Power Plants applications.