Despite the proven benefits of combined heat and power (CHP) and recently introduced subsidies to support it, CHP adoption has not met its targets. One of the possible reasons for this is risk from uncertain electricity and gas prices. To gain insights into the risk management of a CHP unit, we develop a multi-stage stochastic mean-risk optimisation model for the medium-term management of a distributed generation system with a gas-fired microturbine with heat recovery and a boiler. The model adopts the perspective of a large consumer that procures gas (for on-site generation) and electricity (for consumption) on the spot and futures markets. The consumer's risk aversion is incorporated into the model through the conditional value-at-risk (CVaR) measure. We show that CHP not only decreases the consumer's expected cost and risk exposure by 10% each but also improves expected energy efficiency by 4 percentage points and decreases expected CO2 emissions by 16%. The risk exposure can be further mitigated through the use of financial contracts.