Abstract-In this paper, a pricing model is proposed for cointegrated commodities extending Larsson model. The futures formulae have been derived considering a linear combination of a Brownian motion and an Ornstein-Uhlenbeck process describing the co-integration relationship of the different futures prices commodities. Tests have been performed with a non-constant volatility in order to fit better the real behavior of the volatility. The model has been applied to energy commodities (gas, CO2, energy) and soft commodities (corn, wheat). Results show that first, the model can be used with different kind of commodities at the cost of a proper parameters calibration and in second, using a non-constant volatility leads to more accurate short term prices, which provides better evaluation of Value-at-Risk and more generally improves the risk management.