Industrial cogeneration facilities mainly use steam and gas turbines. In a pulp mill, it is often difficult to understand which operating conditions in the steam plant maximize profitability, given the complexity of steam equipment operating envelopes, variability in energy markets and details of electricity purchase agreements. In recent years, CanmetENERGY (NRCan) has developed the COGEN software for the optimization of industrial cogeneration systems. It has also worked on the optimization of cogeneration systems in several Canadian kraft and TMP mills, in partnership with FPInnovations. This article summarizes some key lessons learned from this experience. Optimization at the supervisory level requires a precise understanding of the fuel and steam paths involved in a particular action in the control room. Understanding profitability of path combinations requires a careful distinction of fixed and variable costs related to equipment performance curves, maintenance, ash handling, demineralized water make-up, fuel transport, electricity transmission, etc. Multi-periodicity, often induced by contractual limits to cumulative fuel and electricity flows, adds complexity. In general, pulp mills can increase their profitability by $1-2 million per year by improving decision-making in the control room of the steam plant. In some multiperiod cases, numerical optimization can add $300,000-$400,000 per year.